From 57acf569f65ce516146b3f9c3fe6e1fe40d5869a Mon Sep 17 00:00:00 2001 From: Choy Rim Date: Wed, 10 May 2023 22:35:50 -0400 Subject: [PATCH] fix ref to dair-ai/emotion dataset in 02_classification.ipynb --- 02_classification.ipynb | 7414 +++++++++++++++++++-------------------- 1 file changed, 3707 insertions(+), 3707 deletions(-) diff --git a/02_classification.ipynb b/02_classification.ipynb index c56b56e..b0259c1 100644 --- a/02_classification.ipynb +++ b/02_classification.ipynb @@ -1,3707 +1,3707 @@ -{ - "cells": [ - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# Uncomment and run this cell if you're on Colab or Kaggle\n", - "# !git clone https://github.com/nlp-with-transformers/notebooks.git\n", - "# %cd notebooks\n", - "# from install import *\n", - "# install_requirements(is_chapter2=True)" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# hide\n", - "from utils import *\n", - "setup_chapter()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "# Text Classification" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Text classification is one of the most common tasks in NLP; it can be used for a broad range of applications, such as tagging customer feedback into categories or routing support tickets according to their language. Chances are that your email program's spam filter is using text classification to protect your inbox from a deluge of unwanted junk!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Another common type of text classification is sentiment analysis, which (as we saw in <>) aims to identify the polarity of a given text. For example, a company like Tesla might analyze Twitter posts like the one in <> to determine whether people like its new car roofs or not." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\"Tesla" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Now imagine that you are a data scientist who needs to build a system that can automatically identify emotional states such as \"anger\" or \"joy\" that people express about your company's product on Twitter. In this chapter, we'll tackle this task using a variant of BERT called DistilBERT.footnote:[V. Sanh et al., [\"DistilBERT, a Distilled Version of BERT: Smaller, Faster, Cheaper and Lighter\"](https://arxiv.org/abs/1910.01108), (2019).] The main advantage of this model is that it achieves comparable performance to BERT, while being significantly smaller and more efficient. This enables us to train a classifier in a few minutes, and if you want to train a larger BERT model you can simply change the checkpoint of the pretrained model. A _checkpoint_ corresponds to the set of weights that are loaded into a given transformer architecture.\n", - "\n", - "This will also be our first encounter with three of the core libraries from the Hugging Face ecosystem: image:images/logo.png[hf,13,13] Datasets, image:images/logo.png[hf,13,13] Tokenizers, and image:images/logo.png[hf,13,13] Transformers. As shown in <>, these libraries will allow us to quickly go from raw text to a fine-tuned model that can be used for inference on new tweets. So, in the spirit of Optimus Prime, let's dive in, \"transform, and roll out!\"footnote:[Optimus Prime is the leader of a race of robots in the popular Transformers franchise for children (and for those who are young at heart!).]" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\"Hugging" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## The Dataset" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "To build our emotion detector we'll use a great dataset from an article that explored how emotions are represented in English Twitter messages.footnote:[E. Saravia et al., \"CARER: Contextualized Affect Representations for Emotion Recognition,\" _Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing_ (Oct–Nov 2018): 3687–3697, http://dx.doi.org/10.18653/v1/D18-1404.] Unlike most sentiment analysis datasets that involve just \"positive\" and \"negative\" polarities, this dataset contains six basic emotions: anger, disgust, fear, joy, sadness, and surprise. Given a tweet, our task will be to train a model that can classify it into one of these emotions." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### A First Look at Hugging Face Datasets" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We will use image:images/logo.png[hf,13,13] Datasets to download the data from the [Hugging Face Hub](https://huggingface.co/datasets). We can use the `list_datasets()` function to see what datasets are available on the Hub:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "There are 1753 datasets currently available on the Hub\n", - "The first 10 are: ['acronym_identification', 'ade_corpus_v2', 'adversarial_qa',\n", - "'aeslc', 'afrikaans_ner_corpus', 'ag_news', 'ai2_arc', 'air_dialogue',\n", - "'ajgt_twitter_ar', 'allegro_reviews']\n" - ] - } - ], - "source": [ - "from datasets import list_datasets\n", - "\n", - "all_datasets = list_datasets()\n", - "print(f\"There are {len(all_datasets)} datasets currently available on the Hub\")\n", - "print(f\"The first 10 are: {all_datasets[:10]}\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We see that each dataset is given a name, so let's load the `emotion` dataset with the `load_dataset()` function:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/vnd.jupyter.widget-view+json": { - "model_id": "1454a965e1f5435fbf369734d6608857", - "version_major": 2, - "version_minor": 0 - }, - "text/plain": [ - "Downloading: 0%| | 0.00/1.66k [00:00>." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "```asciidoc\n", - "[[dataset-loading]]\n", - ".How to load datasets in various formats\n", - "[options=\"header\"]\n", - "|======\n", - "| Data format | Loading script | Example\n", - "| CSV | `csv` | `load_dataset(\"csv\", data_files=\"my_file.csv\")` \n", - "| Text | `text` | `load_dataset(\"text\", data_files=\"my_file.txt\")` \n", - "| JSON | `json` | `load_dataset(\"json\", data_files=\"my_file.jsonl\")`\n", - "|======\n", - "```" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As you can see for each data format, we just need to pass the relevant loading script to the `load_dataset()` function, along with a `data_files` argument that specifies the path or URL to one or more files. For example, the source files for the `emotion` dataset are actually hosted on Dropbox, so an alternative way to load the dataset is to first download one of the splits:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "--2023-02-14 19:24:47-- https://huggingface.co/datasets/transformersbook/emotion-train-split/raw/main/train.txt\n", - "Resolving huggingface.co (huggingface.co)... 54.235.118.239, 3.231.67.228, 2600:1f18:147f:e850:e203:c458:10cd:fc3c, ...\n", - "Connecting to huggingface.co (huggingface.co)|54.235.118.239|:443... connected.\n", - "HTTP request sent, awaiting response... 200 OK\n", - "Length: 1658616 (1,6M) [text/plain]\n", - "Saving to: ‘train.txt’\n", - "\n", - "train.txt 100%[===================>] 1,58M 1,06MB/s in 1,5s \n", - "\n", - "2023-02-14 19:24:49 (1,06 MB/s) - ‘train.txt’ saved [1658616/1658616]\n", - "\n" - ] - } - ], - "source": [ - "#hide_output\n", - "# The original URL used in the book is no longer available, so we use a different one\n", - "dataset_url = \"https://huggingface.co/datasets/transformersbook/emotion-train-split/raw/main/train.txt\"\n", - "!wget {dataset_url}" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "If you’re wondering why there’s a `!` character in the preceding shell command, that’s because we’re running the commands in a Jupyter notebook. Simply remove the prefix if you want to download and unzip the dataset within a terminal. Now, if we peek at the first row of the _train.txt_ file:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "i didnt feel humiliated;sadness\n" - ] - } - ], - "source": [ - "!head -n 1 train.txt" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "we can see that here are no column headers and each tweet and emotion are separated by a semicolon. Nevertheless, this is quite similar to a CSV file, so we can load the dataset locally by using the `csv` script and pointing the `data_files` argument to the _train.txt_ file:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "Using custom data configuration default-dd8fa13a78374240\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Downloading and preparing dataset csv/default to /Users/lewtun/.cache/huggingface/datasets/csv/default-dd8fa13a78374240/0.0.0/bf68a4c4aefa545d0712b2fcbb1b327f905bbe2f6425fbc5e8c25234acb9e14a...\n" - ] - }, - { - "data": { - "application/vnd.jupyter.widget-view+json": { - "model_id": "946eefc885d64620a0a05968dcd939f3", - "version_major": 2, - "version_minor": 0 - }, - "text/plain": [ - " 0%| | 0/1 [00:00\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
textlabel
0i didnt feel humiliated0
1i can go from feeling so hopeless to so damned...0
2im grabbing a minute to post i feel greedy wrong3
3i am ever feeling nostalgic about the fireplac...2
4i am feeling grouchy3
\n", - "" - ], - "text/plain": [ - " text label\n", - "0 i didnt feel humiliated 0\n", - "1 i can go from feeling so hopeless to so damned... 0\n", - "2 im grabbing a minute to post i feel greedy wrong 3\n", - "3 i am ever feeling nostalgic about the fireplac... 2\n", - "4 i am feeling grouchy 3" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "import pandas as pd\n", - "\n", - "emotions.set_format(type=\"pandas\")\n", - "df = emotions[\"train\"][:]\n", - "df.head()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As you can see, the column headers have been preserved and the first few rows match our previous views of the data. However, the labels are represented as integers, so let's use the `int2str()` method of the `label` feature to create a new column in our `DataFrame` with the corresponding label names:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
textlabellabel_name
0i didnt feel humiliated0sadness
1i can go from feeling so hopeless to so damned...0sadness
2im grabbing a minute to post i feel greedy wrong3anger
3i am ever feeling nostalgic about the fireplac...2love
4i am feeling grouchy3anger
\n", - "
" - ], - "text/plain": [ - " text label label_name\n", - "0 i didnt feel humiliated 0 sadness\n", - "1 i can go from feeling so hopeless to so damned... 0 sadness\n", - "2 im grabbing a minute to post i feel greedy wrong 3 anger\n", - "3 i am ever feeling nostalgic about the fireplac... 2 love\n", - "4 i am feeling grouchy 3 anger" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "def label_int2str(row):\n", - " return emotions[\"train\"].features[\"label\"].int2str(row)\n", - "\n", - "df[\"label_name\"] = df[\"label\"].apply(label_int2str)\n", - "df.head()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Before diving into building a classifier, let's take a closer look at the dataset. As Andrej Karpathy notes in his famous blog post [\"A Recipe for Training Neural Networks\"](https://karpathy.github.io/2019/04/25/recipe), becoming \"one with the data\" is an essential step for training great models!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Looking at the Class Distribution" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Whenever you are working on text classification problems, it is a good idea to examine the distribution of examples across the classes. A dataset with a skewed class distribution might require a different treatment in terms of the training loss and evaluation metrics than a balanced one. \n", - "\n", - "With Pandas and Matplotlib, we can quickly visualize the class distribution as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "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\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-29T16:17:44.966027\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": { - "needs_background": "light" - }, - "output_type": "display_data" - } - ], - "source": [ - "import matplotlib.pyplot as plt\n", - "\n", - "df[\"label_name\"].value_counts(ascending=True).plot.barh()\n", - "plt.title(\"Frequency of Classes\")\n", - "plt.show()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In this case, we can see that the dataset is heavily imbalanced; the `joy` and `sadness` classes appear frequently, whereas `love` and `surprise` are about 5–10 times rarer. There are several ways to deal with imbalanced data, including:\n", - "\n", - "* Randomly oversample the minority class.\n", - "* Randomly undersample the majority class.\n", - "* Gather more labeled data from the underrepresented classes." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "To keep things simple in this chapter, we'll work with the raw, unbalanced class frequencies. If you want to learn more about these sampling techniques, we recommend checking out the [Imbalanced-learn library](https://imbalanced-learn.org/stable/). Just make sure that you don't apply sampling methods _before_ creating your train/test splits, or you'll get plenty of leakage between them!\n", - "\n", - "Now that we've looked at the classes, let's take a look at the tweets themselves." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### How Long Are Our Tweets?" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Transformer models have a maximum input sequence length that is referred to as the _maximum context size_. For applications using DistilBERT, the maximum context size is 512 tokens, which amounts to a few paragraphs of text. As we'll see in the next section, a token is an atomic piece of text; for now, we'll treat a token as a single word. We can get a rough estimate of tweet lengths per emotion by looking at the distribution of words per tweet:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "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\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-29T16:18:00.302888\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": { - "needs_background": "light" - }, - "output_type": "display_data" - } - ], - "source": [ - "df[\"Words Per Tweet\"] = df[\"text\"].str.split().apply(len)\n", - "df.boxplot(\"Words Per Tweet\", by=\"label_name\", grid=False, showfliers=False,\n", - " color=\"black\")\n", - "plt.suptitle(\"\")\n", - "plt.xlabel(\"\")\n", - "plt.show()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "From the plot we see that for each emotion, most tweets are around 15 words long and the longest tweets are well below DistilBERT's maximum context size. Texts that are longer than a model's context size need to be truncated, which can lead to a loss in performance if the truncated text contains crucial information; in this case, it looks like that won't be an issue. " - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Let's now figure out how we can convert these raw texts into a format suitable for image:images/logo.png[hf,13,13] Transformers! While we're at it, let's also reset the output format of our dataset since we don't need the `DataFrame` format anymore: " - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "emotions.reset_format()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## From Text to Tokens" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Transformer models like DistilBERT cannot receive raw strings as input; instead, they assume the text has been _tokenized_ and _encoded_ as numerical vectors. Tokenization is the step of breaking down a string into the atomic units used in the model. There are several tokenization strategies one can adopt, and the optimal splitting of words into subunits is usually learned from the corpus. Before looking at the tokenizer used for DistilBERT, let's consider two extreme cases: _character_ and _word_ tokenization." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Character Tokenization" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The simplest tokenization scheme is to feed each character individually to the model. In Python, `str` objects are really arrays under the hood, which allows us to quickly implement character-level tokenization with just one line of code:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "['T', 'o', 'k', 'e', 'n', 'i', 'z', 'i', 'n', 'g', ' ', 't', 'e', 'x', 't', ' ',\n", - "'i', 's', ' ', 'a', ' ', 'c', 'o', 'r', 'e', ' ', 't', 'a', 's', 'k', ' ', 'o',\n", - "'f', ' ', 'N', 'L', 'P', '.']\n" - ] - } - ], - "source": [ - "text = \"Tokenizing text is a core task of NLP.\"\n", - "tokenized_text = list(text)\n", - "print(tokenized_text)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This is a good start, but we're not done yet. Our model expects each character to be converted to an integer, a process sometimes called _numericalization_. One simple way to do this is by encoding each unique token (which are characters in this case) with a unique integer:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "{' ': 0, '.': 1, 'L': 2, 'N': 3, 'P': 4, 'T': 5, 'a': 6, 'c': 7, 'e': 8, 'f': 9,\n", - "'g': 10, 'i': 11, 'k': 12, 'n': 13, 'o': 14, 'r': 15, 's': 16, 't': 17, 'x': 18,\n", - "'z': 19}\n" - ] - } - ], - "source": [ - "token2idx = {ch: idx for idx, ch in enumerate(sorted(set(tokenized_text)))}\n", - "print(token2idx)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This gives us a mapping from each character in our vocabulary to a unique integer. We can now use `token2idx` to transform the tokenized text to a list of integers:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "[5, 14, 12, 8, 13, 11, 19, 11, 13, 10, 0, 17, 8, 18, 17, 0, 11, 16, 0, 6, 0, 7,\n", - "14, 15, 8, 0, 17, 6, 16, 12, 0, 14, 9, 0, 3, 2, 4, 1]\n" - ] - } - ], - "source": [ - "input_ids = [token2idx[token] for token in tokenized_text]\n", - "print(input_ids)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Each token has now been mapped to a unique numerical identifier (hence the name `input_ids`). The last step is to convert `input_ids` to a 2D tensor of one-hot vectors. One-hot vectors are frequently used in machine learning to encode categorical data, which can be either ordinal or nominal. For example, suppose we wanted to encode the names of characters in the _Transformers_ TV series. One way to do this would be to map each name to a unique ID, as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
NameLabel ID
0Bumblebee0
1Optimus Prime1
2Megatron2
\n", - "
" - ], - "text/plain": [ - " Name Label ID\n", - "0 Bumblebee 0\n", - "1 Optimus Prime 1\n", - "2 Megatron 2" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "categorical_df = pd.DataFrame(\n", - " {\"Name\": [\"Bumblebee\", \"Optimus Prime\", \"Megatron\"], \"Label ID\": [0,1,2]})\n", - "categorical_df" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The problem with this approach is that it creates a fictitious ordering between the names, and neural networks are _really_ good at learning these kinds of relationships. So instead, we can create a new column for each category and assign a 1 where the category is true, and a 0 otherwise. In Pandas, this can be implemented with the `get_dummies()` function as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
BumblebeeMegatronOptimus Prime
0100
1001
2010
\n", - "
" - ], - "text/plain": [ - " Bumblebee Megatron Optimus Prime\n", - "0 1 0 0\n", - "1 0 0 1\n", - "2 0 1 0" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "pd.get_dummies(categorical_df[\"Name\"])" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The rows of this `DataFrame` are the one-hot vectors, which have a single \"hot\" entry with a 1 and 0s everywhere else. Now, looking at our `input_ids`, we have a similar problem: the elements create an ordinal scale. This means that adding or subtracting two IDs is a meaningless operation, since the result is a new ID that represents another random token.\n", - "\n", - "On the other hand, the result of adding two one-hot encodings can easily be interpreted: the two entries that are \"hot\" indicate that the corresponding tokens co-occur. We can create the one-hot encodings in PyTorch by converting `input_ids` to a tensor and applying the `one_hot()` function as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "torch.Size([38, 20])" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "import torch\n", - "import torch.nn.functional as F\n", - "\n", - "input_ids = torch.tensor(input_ids)\n", - "one_hot_encodings = F.one_hot(input_ids, num_classes=len(token2idx))\n", - "one_hot_encodings.shape" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "For each of the 38 input tokens we now have a one-hot vector with 20 dimensions, since our vocabulary consists of 20 unique characters." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "> Warning: It's important to always set `num_classes` in the `one_hot()` function because otherwise the one-hot vectors may end up being shorter than the length of the vocabulary (and need to be padded with zeros manually). In TensorFlow, the equivalent function is `tf.one_hot()`, where the `depth` argument plays the role of `num_classes`." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "By examining the first vector, we can verify that a 1 appears in the location indicated by `input_ids[0]`:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Token: T\n", - "Tensor index: 5\n", - "One-hot: tensor([0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0])\n" - ] - } - ], - "source": [ - "print(f\"Token: {tokenized_text[0]}\")\n", - "print(f\"Tensor index: {input_ids[0]}\")\n", - "print(f\"One-hot: {one_hot_encodings[0]}\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "From our simple example we can see that character-level tokenization ignores any structure in the text and treats the whole string as a stream of characters. Although this helps deal with misspellings and rare words, the main drawback is that linguistic structures such as words need to be _learned_ from the data. This requires significant compute, memory, and data. For this reason, character tokenization is rarely used in practice. Instead, some structure of the text is preserved during the tokenization step. _Word tokenization_ is a straightforward approach to achieve this, so let's take a look at how it works." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Word Tokenization" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Instead of splitting the text into characters, we can split it into words and map each word to an integer. Using words from the outset enables the model to skip the step of learning words from characters, and thereby reduces the complexity of the training process." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "One simple class of word tokenizers uses whitespace to tokenize the text. We can do this by applying Python's `split()` function directly on the raw text (just like we did to measure the tweet lengths):" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "['Tokenizing', 'text', 'is', 'a', 'core', 'task', 'of', 'NLP.']\n" - ] - } - ], - "source": [ - "tokenized_text = text.split()\n", - "print(tokenized_text)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "From here we can take the same steps we took for the character tokenizer to map each word to an ID. However, we can already see one potential problem with this tokenization scheme: punctuation is not accounted for, so `NLP.` is treated as a single token. Given that words can include declinations, conjugations, or misspellings, the size of the vocabulary can easily grow into the millions! \n", - "\n", - "\n", - "> note: Some word tokenizers have extra rules for punctuation. One can also apply stemming or lemmatization, which normalizes words to their stem (e.g., \"great\", \"greater\", and \"greatest\" all become \"great\"), at the expense of losing some information in the text. " - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Having a large vocabulary is a problem because it requires neural networks to have an enormous number of parameters. To illustrate this, suppose we have 1 million unique words and want to compress the 1-million-dimensional input vectors to 1-thousand-dimensional vectors in the first layer of our neural network. This is a standard step in most NLP architectures, and the resulting weight matrix of this first layer would contain 1 million $\\times$ 1 thousand = 1 billion weights. This is already comparable to the largest GPT-2 model,footnote:[GPT-2 is the successor of GPT, and it captivated the public's attention with its impressive ability to generate realistic text. We'll explore GPT-2 in detail in <>.] which has around 1.5 billion parameters in total!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Naturally, we want to avoid being so wasteful with our model parameters since models are expensive to train, and larger models are more difficult to maintain. A common approach is to limit the vocabulary and discard rare words by considering, say, the 100,000 most common words in the corpus. Words that are not part of the vocabulary are classified as \"unknown\" and mapped to a shared `UNK` token. This means that we lose some potentially important information in the process of word tokenization, since the model has no information about words associated with `UNK`.\n", - "\n", - "Wouldn't it be nice if there was a compromise between character and word tokenization that preserved all the input information _and_ some of the input structure? There is: _subword tokenization_." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Subword Tokenization" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The basic idea behind subword tokenization is to combine the best aspects of character and word tokenization. On the one hand, we want to split rare words into smaller units to allow the model to deal with complex words and misspellings. On the other hand, we want to keep frequent words as unique entities so that we can keep the length of our inputs to a manageable size. The main distinguishing feature of subword tokenization (as well as word tokenization) is that it is _learned_ from the pretraining corpus using a mix of statistical rules and algorithms.\n", - "\n", - "There are several subword tokenization algorithms that are commonly used in NLP, but let's start with WordPiece,footnote:[M. Schuster and K. Nakajima, \"Japanese and Korean Voice Search,\" _2012 IEEE International Conference on Acoustics, Speech and Signal Processing_ (2012): 5149–5152, https://doi.org/10.1109/ICASSP.2012.6289079.] which is used by the BERT and DistilBERT tokenizers. The easiest way to understand how WordPiece works is to see it in action. image:images/logo.png[hf,13,13] Transformers provides a convenient `AutoTokenizer` class that allows you to quickly load the tokenizer associated with a pretrained model—we just call its `from_pretrained()` method, providing the ID of a model on the Hub or a local file path. Let's start by loading the tokenizer for DistilBERT:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# hide_output\n", - "from transformers import AutoTokenizer\n", - "\n", - "model_ckpt = \"distilbert-base-uncased\"\n", - "tokenizer = AutoTokenizer.from_pretrained(model_ckpt)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The `AutoTokenizer` class belongs to a larger set of [\"auto\" classes](https://huggingface.co/docs/transformers/model_doc/auto) whose job is to automatically retrieve the model's configuration, pretrained weights, or vocabulary from the name of the checkpoint. This allows you to quickly switch between models, but if you wish to load the specific class manually you can do so as well. For example, we could have loaded the DistilBERT tokenizer as follows:\n", - "\n", - "```python\n", - "from transformers import DistilBertTokenizer\n", - "\n", - "distilbert_tokenizer = DistilBertTokenizer.from_pretrained(model_ckpt)\n", - "```" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "> note: When you run the `AutoTokenizer.from_pretrained()` method for the first time you will see a progress bar that shows which parameters of the pretrained tokenizer are loaded from the Hugging Face Hub. When you run the code a second time, it will load the tokenizer from the cache, usually located at _~/.cache/huggingface/_." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Let's examine how this tokenizer works by feeding it our simple \"Tokenizing text is a core task of NLP.\" example text:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "{'input_ids': [101, 19204, 6026, 3793, 2003, 1037, 4563, 4708, 1997, 17953,\n", - "2361, 1012, 102], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}\n" - ] - } - ], - "source": [ - "encoded_text = tokenizer(text)\n", - "print(encoded_text)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Just like we saw with character tokenization, we can see that the words have been mapped to unique integers in the `input_ids` field. We'll discuss the role of the `attention_mask` field in the next section. Now that we have the `input_ids`, we can convert them back into tokens by using the tokenizer's `convert_ids_to_tokens()` method:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "['[CLS]', 'token', '##izing', 'text', 'is', 'a', 'core', 'task', 'of', 'nl',\n", - "'##p', '.', '[SEP]']\n" - ] - } - ], - "source": [ - "tokens = tokenizer.convert_ids_to_tokens(encoded_text.input_ids)\n", - "print(tokens)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We can observe three things here. First, some special `[CLS]` and `[SEP]` tokens have been added to the start and end of the sequence. These tokens differ from model to model, but their main role is to indicate the start and end of a sequence. Second, the tokens have each been lowercased, which is a feature of this particular checkpoint. Finally, we can see that \"tokenizing\" and \"NLP\" have been split into two tokens, which makes sense since they are not common words. The `##` prefix in `##izing` and `##p` means that the preceding string is not whitespace; any token with this prefix should be merged with the previous token when you convert the tokens back to a string. The `AutoTokenizer` class has a `convert_tokens_to_string()` method for doing just that, so let's apply it to our tokens:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "[CLS] tokenizing text is a core task of nlp. [SEP]\n" - ] - } - ], - "source": [ - "print(tokenizer.convert_tokens_to_string(tokens))" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The `AutoTokenizer` class also has several attributes that provide information about the tokenizer. For example, we can inspect the vocabulary size:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "30522" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "tokenizer.vocab_size" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "and the corresponding model's maximum context size:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "512" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "tokenizer.model_max_length" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Another interesting attribute to know about is the names of the fields that the model expects in its forward pass:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "['input_ids', 'attention_mask']" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "tokenizer.model_input_names" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Now that we have a basic understanding of the tokenization process for a single string, let's see how we can tokenize the whole dataset!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "> warning: When using pretrained models, it is _really_ important to make sure that you use the same tokenizer that the model was trained with. From the model's perspective, switching the tokenizer is like shuffling the vocabulary. If everyone around you started swapping random words like \"house\" for \"cat,\" you'd have a hard time understanding what was going on too!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Tokenizing the Whole Dataset" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "To tokenize the whole corpus, we'll use the `map()` method of our `DatasetDict` object. We'll encounter this method many times throughout this book, as it provides a convenient way to apply a processing function to each element in a dataset. As we'll soon see, the `map()` method can also be used to create new rows and columns.\n", - "\n", - "To get started, the first thing we need is a processing function to tokenize our examples with:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "def tokenize(batch):\n", - " return tokenizer(batch[\"text\"], padding=True, truncation=True)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This function applies the tokenizer to a batch of examples; `padding=True` will pad the examples with zeros to the size of the longest one in a batch, and `truncation=True` will truncate the examples to the model's maximum context size. To see `tokenize()` in action, let's pass a batch of two examples from the training set:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "{'input_ids': [[101, 1045, 2134, 2102, 2514, 26608, 102, 0, 0, 0, 0, 0, 0, 0, 0,\n", - "0, 0, 0, 0, 0, 0, 0, 0], [101, 1045, 2064, 2175, 2013, 3110, 2061, 20625, 2000,\n", - "2061, 9636, 17772, 2074, 2013, 2108, 2105, 2619, 2040, 14977, 1998, 2003, 8300,\n", - "102]], 'attention_mask': [[1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n", - "0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n", - "1, 1]]}\n" - ] - } - ], - "source": [ - "print(tokenize(emotions[\"train\"][:2]))" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Here we can see the result of padding: the first element of `input_ids` is shorter than the second, so zeros have been added to that element to make them the same length. These zeros have a corresponding `[PAD]` token in the vocabulary, and the set of special tokens also includes the `[CLS]` and `[SEP]` tokens that we encountered earlier:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
Special Token[PAD][UNK][CLS][SEP][MASK]
Special Token ID0100101102103
" - ], - "text/plain": [ - "" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "#hide_input\n", - "tokens2ids = list(zip(tokenizer.all_special_tokens, tokenizer.all_special_ids))\n", - "data = sorted(tokens2ids, key=lambda x : x[-1])\n", - "df = pd.DataFrame(data, columns=[\"Special Token\", \"Special Token ID\"])\n", - "df.T" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Also note that in addition to returning the encoded tweets as `input_ids`, the tokenizer returns a list of `attention_mask` arrays. This is because we do not want the model to get confused by the additional padding tokens: the attention mask allows the model to ignore the padded parts of the input. <> provides a visual explanation of how the input IDs and attention masks are padded.\n", - "\n", - "\"attention-mask\" " - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Once we've defined a processing function, we can apply it across all the splits in the corpus in a single line of code:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# hide_output\n", - "emotions_encoded = emotions.map(tokenize, batched=True, batch_size=None)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "By default, the `map()` method operates individually on every example in the corpus, so setting `batched=True` will encode the tweets in batches. Because we've set `batch_size=None`, our `tokenize()` function will be applied on the full dataset as a single batch. This ensures that the input tensors and attention masks have the same shape globally, and we can see that this operation has added new `input_ids` and `attention_mask` columns to the dataset:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "['attention_mask', 'input_ids', 'label', 'text']\n" - ] - } - ], - "source": [ - "print(emotions_encoded[\"train\"].column_names)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "> Note: In later chapters, we'll see how _data collators_ can be used to dynamically pad the tensors in each batch. Padding globally will come in handy in the next section, where we extract a feature matrix from the whole corpus." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## Training a Text Classifier" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As discussed in <>, models like DistilBERT are pretrained to predict masked words in a sequence of text. However, we can't use these language models directly for text classification; we need to modify them slightly. To understand what modifications are necessary, let's take a look at the architecture of an encoder-based model like DistilBERT, which is depicted in <>. " - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\"encoder-classifier\"" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "First, the text is tokenized and represented as one-hot vectors called _token encodings_. The size of the tokenizer vocabulary determines the dimension of the token encodings, and it usually consists of 20k–200k unique tokens. Next, these token encodings are converted to _token embeddings_, which are vectors living in a lower-dimensional space. The token embeddings are then passed through the encoder block layers to yield a _hidden state_ for each input token. For the pretraining objective of language modeling,⁠footnote:[In the case of DistilBERT, it's guessing the masked tokens.] each hidden state is fed to a layer that predicts the masked input tokens. For the classification task, we replace the language modeling layer with a classification layer." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "> note: In practice, PyTorch skips the step of creating one-hot vectors for token encodings because multiplying a matrix with a one-hot vector is the same as selecting a column from the matrix. This can be done directly by getting the column with the token ID from the matrix. We'll see this in <> when we use the `nn.Embedding` class." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We have two options to train such a model on our Twitter dataset:\n", - "\n", - "- _Feature extraction_:: We use the hidden states as features and just train a classifier on them, without modifying the pretrained model.\n", - "- _Fine-tuning_:: We train the whole model end-to-end, which also updates the parameters of the pretrained model. \n", - "\n", - "In the following sections we explore both options for DistilBERT and examine their trade-offs. " - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Transformers as Feature Extractors" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - "Using a transformer as a feature extractor is fairly simple. As shown in <>, we freeze the body's weights during training and use the hidden states as features for the classifier. The advantage of this approach is that we can quickly train a small or shallow model. Such a model could be a neural classification layer or a method that does not rely on gradients, such as a random forest. This method is especially convenient if GPUs are unavailable, since the hidden states only need to be precomputed once." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\"encoder-features\"" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Using pretrained models" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - "We will use another convenient auto class from image:images/logo.png[hf,13,13] Transformers called `AutoModel`. Similar to the `AutoTokenizer` class, `AutoModel` has a `from_pretrained()` method to load the weights of a pretrained model. Let's use this method to load the DistilBERT checkpoint:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# hide_output\n", - "from transformers import AutoModel\n", - "\n", - "model_ckpt = \"distilbert-base-uncased\"\n", - "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n", - "model = AutoModel.from_pretrained(model_ckpt).to(device)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Here we've used PyTorch to check whether a GPU is available or not, and then chained the PyTorch `nn.Module.to()` method to the model loader. This ensures that the model will run on the GPU if we have one. If not, the model will run on the CPU, which can be considerably slower." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The `AutoModel` class converts the token encodings to embeddings, and then feeds them through the encoder stack to return the hidden states. Let's take a look at how we can extract these states from our corpus." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Sidebar: Interoperability Between Frameworks" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Although the code in this book is mostly written in PyTorch, image:images/logo.png[hf,13,13] Transformers provides tight interoperability with TensorFlow and JAX. This means that you only need to change a few lines of code to load a pretrained model in your favorite deep learning framework! For example, we can load DistilBERT in TensorFlow by using the `TFAutoModel` class as follows: " - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2021-10-23 17:03:51.654626: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory\n", - "2021-10-23 17:03:51.654664: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1835] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n", - "Skipping registering GPU devices...\n", - "2021-10-23 17:03:51.655491: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA\n", - "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n", - "2021-10-23 17:03:51.680031: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n" - ] - } - ], - "source": [ - "#hide_output\n", - "from transformers import TFAutoModel\n", - "\n", - "tf_model = TFAutoModel.from_pretrained(model_ckpt)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This interoperability is especially useful when a model is only released in one framework, but you'd like to use it in another. For example, the [XLM-RoBERTa model](https://huggingface.co/xlm-roberta-base) that we'll encounter in <> only has PyTorch weights, so if you try to load it in TensorFlow as we did before:\n", - "\n", - "```python\n", - "tf_xlmr = TFAutoModel.from_pretrained(\"xlm-roberta-base\")\n", - "```\n", - "\n", - "you'll get an error. In these cases, you can specify a `from_pt=True` argument to the `TfAutoModel.from_pretrained()` function, and the library will automatically download and convert the PyTorch weights for you:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "tf_xlmr = TFAutoModel.from_pretrained(\"xlm-roberta-base\", from_pt=True)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As you can see, it is very simple to switch between frameworks in image:images/logo.png[hf,13,13] Transformers! In most cases, you can just add a \"TF\" prefix to the classes and you'll get the equivalent TensorFlow 2.0 classes. When we use the `\"pt\"` string (e.g., in the following section), which is short for PyTorch, just replace it with \"`tf\"`, which is short for TensorFlow." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### End sidebar" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Extracting the last hidden states" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "To warm up, let's retrieve the last hidden states for a single string. The first thing we need to do is encode the string and convert the tokens to PyTorch tensors. This can be done by providing the `return_tensors=\"pt\"` argument to the tokenizer as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Input tensor shape: torch.Size([1, 6])\n" - ] - } - ], - "source": [ - "text = \"this is a test\"\n", - "inputs = tokenizer(text, return_tensors=\"pt\")\n", - "print(f\"Input tensor shape: {inputs['input_ids'].size()}\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As we can see, the resulting tensor has the shape `[batch_size, n_tokens]`. Now that we have the encodings as a tensor, the final step is to place them on the same device as the model and pass the inputs as follows: " - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "BaseModelOutput(last_hidden_state=tensor([[[-0.1565, -0.1862, 0.0528, ...,\n", - "-0.1188, 0.0662, 0.5470],\n", - " [-0.3575, -0.6484, -0.0618, ..., -0.3040, 0.3508, 0.5221],\n", - " [-0.2772, -0.4459, 0.1818, ..., -0.0948, -0.0076, 0.9958],\n", - " [-0.2841, -0.3917, 0.3753, ..., -0.2151, -0.1173, 1.0526],\n", - " [ 0.2661, -0.5094, -0.3180, ..., -0.4203, 0.0144, -0.2149],\n", - " [ 0.9441, 0.0112, -0.4714, ..., 0.1439, -0.7288, -0.1619]]],\n", - " device='cuda:0'), hidden_states=None, attentions=None)\n" - ] - } - ], - "source": [ - "inputs = {k:v.to(device) for k,v in inputs.items()}\n", - "with torch.no_grad():\n", - " outputs = model(**inputs)\n", - "print(outputs)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Here we've used the `torch.no_grad()` context manager to disable the automatic calculation of the gradient. This is useful for inference since it reduces the memory footprint of the computations. Depending on the model configuration, the output can contain several objects, such as the hidden states, losses, or attentions, arranged in a class similar to a `namedtuple` in Python. In our example, the model output is an instance of `BaseModelOutput`, and we can simply access its attributes by name. The current model returns only one attribute, which is the last hidden state, so let's examine its shape:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "torch.Size([1, 6, 768])" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "outputs.last_hidden_state.size()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Looking at the hidden state tensor, we see that it has the shape `[batch_size, n_tokens, hidden_dim]`. In other words, a 768-dimensional vector is returned for each of the 6 input tokens. For classification tasks, it is common practice to just use the hidden state associated with the `[CLS]` token as the input feature. Since this token appears at the start of each sequence, we can extract it by simply indexing into `outputs.last_hidden_state` as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "torch.Size([1, 768])" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "outputs.last_hidden_state[:,0].size()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Now that we know how to get the last hidden state for a single string, let's do the same thing for the whole dataset by creating a new `hidden_state` column that stores all these vectors. As we did with the tokenizer, we'll use the `map()` method of `DatasetDict` to extract all the hidden states in one go. The first thing we need to do is wrap the previous steps in a processing function:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "def extract_hidden_states(batch):\n", - " # Place model inputs on the GPU\n", - " inputs = {k:v.to(device) for k,v in batch.items() \n", - " if k in tokenizer.model_input_names}\n", - " # Extract last hidden states\n", - " with torch.no_grad():\n", - " last_hidden_state = model(**inputs).last_hidden_state\n", - " # Return vector for [CLS] token\n", - " return {\"hidden_state\": last_hidden_state[:,0].cpu().numpy()}" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The only difference between this function and our previous logic is the final step where we place the final hidden state back on the CPU as a NumPy array. The `map()` method requires the processing function to return Python or NumPy objects when we're using batched inputs.\n", - "\n", - "Since our model expects tensors as inputs, the next thing to do is convert the `input_ids` and `attention_mask` columns to the `\"torch\"` format, as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "emotions_encoded.set_format(\"torch\", \n", - " columns=[\"input_ids\", \"attention_mask\", \"label\"])" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We can then go ahead and extract the hidden states across all splits in one go:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/vnd.jupyter.widget-view+json": { - "model_id": "590842bb15bf448cb35e324e87fdadd9", - "version_major": 2, - "version_minor": 0 - }, - "text/plain": [ - " 0%| | 0/16 [00:00\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
XYlabel
04.3580756.1408160
1-3.1345675.3294460
25.1522302.7326433
3-2.5190183.0672502
4-3.3645203.3566133
\n", - "" - ], - "text/plain": [ - " X Y label\n", - "0 4.358075 6.140816 0\n", - "1 -3.134567 5.329446 0\n", - "2 5.152230 2.732643 3\n", - "3 -2.519018 3.067250 2\n", - "4 -3.364520 3.356613 3" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "from umap import UMAP\n", - "from sklearn.preprocessing import MinMaxScaler\n", - "\n", - "# Scale features to [0,1] range\n", - "X_scaled = MinMaxScaler().fit_transform(X_train)\n", - "# Initialize and fit UMAP\n", - "mapper = UMAP(n_components=2, metric=\"cosine\").fit(X_scaled)\n", - "# Create a DataFrame of 2D embeddings\n", - "df_emb = pd.DataFrame(mapper.embedding_, columns=[\"X\", \"Y\"])\n", - "df_emb[\"label\"] = y_train\n", - "df_emb.head()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The result is an array with the same number of training samples, but with only 2 features instead of the 768 we started with! Let's investigate the compressed data a little bit further and plot the density of points for each category separately:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "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\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-23T17:07:41.645879\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "fig, axes = plt.subplots(2, 3, figsize=(7,5))\n", - "axes = axes.flatten()\n", - "cmaps = [\"Greys\", \"Blues\", \"Oranges\", \"Reds\", \"Purples\", \"Greens\"]\n", - "labels = emotions[\"train\"].features[\"label\"].names\n", - "\n", - "for i, (label, cmap) in enumerate(zip(labels, cmaps)):\n", - " df_emb_sub = df_emb.query(f\"label == {i}\")\n", - " axes[i].hexbin(df_emb_sub[\"X\"], df_emb_sub[\"Y\"], cmap=cmap,\n", - " gridsize=20, linewidths=(0,))\n", - " axes[i].set_title(label)\n", - " axes[i].set_xticks([]), axes[i].set_yticks([])\n", - "\n", - "plt.tight_layout()\n", - "plt.show()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - ">note: These are only projections onto a lower-dimensional space. Just because some categories overlap does not mean that they are not separable in the original space. Conversely, if they are separable in the projected space they will be separable in the original space.\n", - "\n", - "From this plot we can see some clear patterns: the negative feelings such as `sadness`, `anger`, and `fear` all occupy similar regions with slightly varying distributions. On the other hand, `joy` and `love` are well separated from the negative emotions and also share a similar space. Finally, `surprise` is scattered all over the place. Although we may have hoped for some separation, this is in no way guaranteed since the model was not trained to know the difference between these emotions. It only learned them implicitly by guessing the masked words in texts.\n", - "\n", - "Now that we've gained some insight into the features of our dataset, let's finally train a model on it!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Training a simple classifier\n" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We've seen that the hidden states are somewhat different between the emotions, although for several of them there is no obvious boundary. Let's use these hidden states to train a logistic regression model with Scikit-Learn. Training such a simple model is fast and does not require a GPU:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/home/lewis/miniconda3/envs/book/lib/python3.7/site-packages/sklearn/linear_model/_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n", - "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n", - "\n", - "Increase the number of iterations (max_iter) or scale the data as shown in:\n", - " https://scikit-learn.org/stable/modules/preprocessing.html\n", - "Please also refer to the documentation for alternative solver options:\n", - " https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n", - " extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n" - ] - }, - { - "data": { - "text/plain": [ - "LogisticRegression()" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "#hide_output\n", - "# We increase `max_iter` to guarantee convergence \n", - "from sklearn.linear_model import LogisticRegression\n", - "\n", - "lr_clf = LogisticRegression(max_iter=3000)\n", - "lr_clf.fit(X_train, y_train)" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "0.6085" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "lr_clf.score(X_valid, y_valid)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Looking at the accuracy, it might appear that our model is just a bit better than random—but since we are dealing with an unbalanced multiclass dataset, it's actually significantly better. We can examine whether our model is any good by comparing it against a simple baseline. In Scikit-Learn there is a `DummyClassifier` that can be used to build a classifier with simple heuristics such as always choosing the majority class or always drawing a random class. In this case the best-performing heuristic is to always choose the most frequent class, which yields an accuracy of about 35%:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "0.352" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "from sklearn.dummy import DummyClassifier\n", - "\n", - "dummy_clf = DummyClassifier(strategy=\"most_frequent\")\n", - "dummy_clf.fit(X_train, y_train)\n", - "dummy_clf.score(X_valid, y_valid)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "So, our simple classifier with DistilBERT embeddings is significantly better than our baseline. We can further investigate the performance of the model by looking at the confusion matrix of the classifier, which tells us the relationship between the true and predicted labels:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "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\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-12-07T11:46:25.597214\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": { - "needs_background": "light" - }, - "output_type": "display_data" - } - ], - "source": [ - "from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix\n", - "\n", - "def plot_confusion_matrix(y_preds, y_true, labels):\n", - " cm = confusion_matrix(y_true, y_preds, normalize=\"true\")\n", - " fig, ax = plt.subplots(figsize=(6, 6))\n", - " disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=labels)\n", - " disp.plot(cmap=\"Blues\", values_format=\".2f\", ax=ax, colorbar=False)\n", - " plt.title(\"Normalized confusion matrix\")\n", - " plt.show()\n", - " \n", - "y_preds = lr_clf.predict(X_valid)\n", - "plot_confusion_matrix(y_preds, y_valid, labels)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We can see that `anger` and `fear` are most often confused with `sadness`, which agrees with the observation we made when visualizing the embeddings. Also, `love` and `surprise` are frequently mistaken for `joy`." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In the next section we will explore the fine-tuning approach, which leads to superior classification performance. It is, however, important to note that doing this requires more computational resources, such as GPUs, that might not be available in your organization. In cases like these, a feature-based approach can be a good compromise between doing traditional machine learning and deep learning." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Fine-Tuning Transformers" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - "Let's now explore what it takes to fine-tune a transformer end-to-end. With the fine-tuning approach we do not use the hidden states as fixed features, but instead train them as shown in <>. This requires the classification head to be differentiable, which is why this method usually uses a neural network for classification.\n", - "\n", - "\"encoder-tuning\"" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Training the hidden states that serve as inputs to the classification model will help us avoid the problem of working with data that may not be well suited for the classification task. Instead, the initial hidden states adapt during training to decrease the model loss and thus increase its performance.\n", - "\n", - "We'll be using the `Trainer` API from image:images/logo.png[hf,13,13] Transformers to simplify the training loop. Let's look at the ingredients we need to set one up!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Loading a pretrained model" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The first thing we need is a pretrained DistilBERT model like the one we used in the feature-based approach. The only slight modification is that we use the `AutoModelForSequenceClassification` model instead of `AutoModel`. The difference is that the `AutoModelForSequenceClassification` model has a classification head on top of the pretrained model outputs, which can be easily trained with the base model. We just need to specify how many labels the model has to predict (six in our case), since this dictates the number of outputs the classification head has:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# hide_output\n", - "from transformers import AutoModelForSequenceClassification\n", - "\n", - "num_labels = 6\n", - "model = (AutoModelForSequenceClassification\n", - " .from_pretrained(model_ckpt, num_labels=num_labels)\n", - " .to(device))" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "You will see a warning that some parts of the model are randomly initialized. This is normal since the classification head has not yet been trained. The next step is to define the metrics that we'll use to evaluate our model's performance during fine-tuning." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Defining the performance metrics" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - "To monitor metrics during training, we need to define a `compute_metrics()` function for the `Trainer`. This function receives an `EvalPrediction` object (which is a named tuple with `predictions` and `label_ids` attributes) and needs to return a dictionary that maps each metric's name to its value. For our application, we'll compute the $F_1$-score and the accuracy of the model as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "from sklearn.metrics import accuracy_score, f1_score\n", - "\n", - "def compute_metrics(pred):\n", - " labels = pred.label_ids\n", - " preds = pred.predictions.argmax(-1)\n", - " f1 = f1_score(labels, preds, average=\"weighted\")\n", - " acc = accuracy_score(labels, preds)\n", - " return {\"accuracy\": acc, \"f1\": f1}" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "With the dataset and metrics ready, we just have two final things to take care of before we define the `Trainer` class:\n", - "\n", - "1. Log in to our account on the Hugging Face Hub. This will allow us to push our fine-tuned model to our account on the Hub and share it with the community.\n", - "2. Define all the hyperparameters for the training run.\n", - "\n", - "We'll tackle these steps in the next section." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Training the model" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "If you're running this code in a Jupyter notebook, you can log in to the Hub with the following helper function:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "from huggingface_hub import notebook_login\n", - "\n", - "notebook_login()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This will display a widget in which you can enter your username and password, or an access token with write privileges. You can find details on how to create access tokens in the [Hub documentation](https://huggingface.co/docs/hub/security#user-access-tokens). If you're working in the terminal, you can log in by running the following command:\n", - "\n", - "```bash\n", - "$ huggingface-cli login\n", - "```" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "To define the training parameters, we use the `TrainingArguments` class. This class stores a lot of information and gives you fine-grained control over the training and evaluation. The most important argument to specify is `output_dir`, which is where all the artifacts from training are stored. Here is an example of `TrainingArguments` in all its glory:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "from transformers import Trainer, TrainingArguments\n", - "\n", - "batch_size = 64\n", - "logging_steps = len(emotions_encoded[\"train\"]) // batch_size\n", - "model_name = f\"{model_ckpt}-finetuned-emotion\"\n", - "training_args = TrainingArguments(output_dir=model_name,\n", - " num_train_epochs=2,\n", - " learning_rate=2e-5,\n", - " per_device_train_batch_size=batch_size,\n", - " per_device_eval_batch_size=batch_size,\n", - " weight_decay=0.01,\n", - " evaluation_strategy=\"epoch\",\n", - " disable_tqdm=False,\n", - " logging_steps=logging_steps,\n", - " push_to_hub=True, \n", - " log_level=\"error\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Here we also set the batch size, learning rate, and number of epochs, and specify to load the best model at the end of the training run. With this final ingredient, we can instantiate and fine-tune our model with the `Trainer`: " - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "\n", - "
\n", - " \n", - " \n", - " [500/500 01:48, Epoch 2/2]\n", - "
\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
EpochTraining LossValidation LossAccuracyF1
10.8409000.3274450.8965000.892285
20.2550000.2204720.9225000.922550

" - ], - "text/plain": [ - "" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "from transformers import Trainer\n", - "\n", - "trainer = Trainer(model=model, args=training_args, \n", - " compute_metrics=compute_metrics,\n", - " train_dataset=emotions_encoded[\"train\"],\n", - " eval_dataset=emotions_encoded[\"validation\"],\n", - " tokenizer=tokenizer)\n", - "trainer.train();" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Looking at the logs, we can see that our model has an $F_1$-score on the validation set of around 92% - this is a significant improvement over the feature-based approach!\n", - "\n", - "We can take a more detailed look at the training metrics by calculating the confusion matrix. To visualize the confusion matrix, we first need to get the predictions on the validation set. The `predict()` method of the `Trainer` class returns several useful objects we can use for evaluation:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "\n", - "

\n", - " \n", - " \n", - " [32/32 00:01]\n", - "
\n", - " " - ], - "text/plain": [ - "" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "# hide_output\n", - "preds_output = trainer.predict(emotions_encoded[\"validation\"])" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "The output of the `predict()` method is a `PredictionOutput` object that contains arrays of `predictions` and `label_ids`, along with the metrics we passed to the trainer. For example, the metrics on the validation set can be accessed as follows:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "{'test_loss': 0.22047173976898193,\n", - " 'test_accuracy': 0.9225,\n", - " 'test_f1': 0.9225500751072866,\n", - " 'test_runtime': 1.6357,\n", - " 'test_samples_per_second': 1222.725,\n", - " 'test_steps_per_second': 19.564}" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "preds_output.metrics" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "It also contains the raw predictions for each class. We can decode the predictions greedily using `np.argmax()`. This yields the predicted labels and has the same format as the labels returned by the Scikit-Learn models in the feature-based approach:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "y_preds = np.argmax(preds_output.predictions, axis=1)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "With the predictions, we can plot the confusion matrix again:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "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\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-12-07T11:59:17.791697\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": { - "needs_background": "light" - }, - "output_type": "display_data" - } - ], - "source": [ - "plot_confusion_matrix(y_preds, y_valid, labels)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "This is much closer to the ideal diagonal confusion matrix. The `love` category is still often confused with `joy`, which seems natural. `surprise` is also frequently mistaken for `joy`, or confused with `fear`. Overall the performance of the model seems quite good, but before we call it a day, let's dive a little deeper into the types of errors our model is likely to make." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### Sidebar: Fine-Tuning with Keras" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "If you are using TensorFlow, it's also possible to fine-tune your models using the Keras API. The main difference from the PyTorch API is that there is no `Trainer` class, since Keras models already provide a built-in `fit()` method. To see how this works, let's first load DistilBERT as a TensorFlow model:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2021-10-29 15:33:36.938811: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory\n", - "2021-10-29 15:33:36.938844: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1835] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n", - "Skipping registering GPU devices...\n", - "2021-10-29 15:33:36.939933: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA\n", - "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n", - "2021-10-29 15:33:36.962642: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n" - ] - } - ], - "source": [ - "#hide_output\n", - "from transformers import TFAutoModelForSequenceClassification\n", - "\n", - "tf_model = (TFAutoModelForSequenceClassification\n", - " .from_pretrained(model_ckpt, num_labels=num_labels))" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Next, we'll convert our datasets into the `tf.data.Dataset` format. Since we have already padded our tokenized inputs, we can do this easily by applying the `to_tf_dataset()` method to `emotions_encoded`:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# The column names to convert to TensorFlow tensors\n", - "tokenizer_columns = tokenizer.model_input_names\n", - "\n", - "tf_train_dataset = emotions_encoded[\"train\"].to_tf_dataset(\n", - " columns=tokenizer_columns, label_cols=[\"label\"], shuffle=True,\n", - " batch_size=batch_size)\n", - "tf_eval_dataset = emotions_encoded[\"validation\"].to_tf_dataset(\n", - " columns=tokenizer_columns, label_cols=[\"label\"], shuffle=False,\n", - " batch_size=batch_size)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Here we've also shuffled the training set, and defined the batch size for it and the validation set. The last thing to do is compile and train the model:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2021-10-29 15:36:00.548707: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Epoch 1/2\n", - "250/250 [==============================] - 478s 2s/step - loss: 0.5379 - sparse_categorical_accuracy: 0.8138 - val_loss: 0.1452 - val_sparse_categorical_accuracy: 0.9430\n", - "Epoch 2/2\n", - "250/250 [==============================] - 471s 2s/step - loss: 0.1424 - sparse_categorical_accuracy: 0.9415 - val_loss: 0.1512 - val_sparse_categorical_accuracy: 0.9335\n" - ] - }, - { - "data": { - "text/plain": [ - "" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "#hide_output\n", - "import tensorflow as tf\n", - "\n", - "tf_model.compile(\n", - " optimizer=tf.keras.optimizers.Adam(learning_rate=5e-5),\n", - " loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n", - " metrics=tf.metrics.SparseCategoricalAccuracy())\n", - "\n", - "tf_model.fit(tf_train_dataset, validation_data=tf_eval_dataset, epochs=2)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### End sidebar" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Error analysis" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Before moving on, we should investigate our model's predictions a little bit further. A simple yet powerful technique is to sort the validation samples by the model loss. When we pass the label during the forward pass, the loss is automatically calculated and returned. Here's a function that returns the loss along with the predicted label:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "from torch.nn.functional import cross_entropy\n", - "\n", - "def forward_pass_with_label(batch):\n", - " # Place all input tensors on the same device as the model\n", - " inputs = {k:v.to(device) for k,v in batch.items() \n", - " if k in tokenizer.model_input_names}\n", - "\n", - " with torch.no_grad():\n", - " output = model(**inputs)\n", - " pred_label = torch.argmax(output.logits, axis=-1)\n", - " loss = cross_entropy(output.logits, batch[\"label\"].to(device), \n", - " reduction=\"none\")\n", - "\n", - " # Place outputs on CPU for compatibility with other dataset columns \n", - " return {\"loss\": loss.cpu().numpy(), \n", - " \"predicted_label\": pred_label.cpu().numpy()}" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Using the `map()` method once more, we can apply this function to get the losses for all the samples:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/vnd.jupyter.widget-view+json": { - "model_id": "6004443ac1344ee18d40c8a90c1178f4", - "version_major": 2, - "version_minor": 0 - }, - "text/plain": [ - " 0%| | 0/125 [00:00\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
textlabelpredicted_labelloss
1801i feel that he was being overshadowed by the s...lovesadness5.704531
1963i called myself pro life and voted for perry w...joysadness5.484461
1870i guess i feel betrayed because i admired him ...joysadness5.434768
882i feel badly about reneging on my commitment t...lovesadness5.257482
1950i as representative of everything thats wrong ...surprisesadness4.827708
1509i guess this is a memoir so it feels like that...joyfear4.713047
1274i am going to several holiday parties and i ca...joysadness4.704955
318i felt ashamed of these feelings and was scare...fearsadness4.656096
1500i guess we would naturally feel a sense of lon...angersadness4.593202
1111im lazy my characters fall into categories of ...joyfear4.311287
\n", - "" - ], - "text/plain": [ - " text label \\\n", - "1801 i feel that he was being overshadowed by the s... love \n", - "1963 i called myself pro life and voted for perry w... joy \n", - "1870 i guess i feel betrayed because i admired him ... joy \n", - "882 i feel badly about reneging on my commitment t... love \n", - "1950 i as representative of everything thats wrong ... surprise \n", - "1509 i guess this is a memoir so it feels like that... joy \n", - "1274 i am going to several holiday parties and i ca... joy \n", - "318 i felt ashamed of these feelings and was scare... fear \n", - "1500 i guess we would naturally feel a sense of lon... anger \n", - "1111 im lazy my characters fall into categories of ... joy \n", - "\n", - " predicted_label loss \n", - "1801 sadness 5.704531 \n", - "1963 sadness 5.484461 \n", - "1870 sadness 5.434768 \n", - "882 sadness 5.257482 \n", - "1950 sadness 4.827708 \n", - "1509 fear 4.713047 \n", - "1274 sadness 4.704955 \n", - "318 sadness 4.656096 \n", - "1500 sadness 4.593202 \n", - "1111 fear 4.311287 " - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "#hide_output\n", - "df_test.sort_values(\"loss\", ascending=False).head(10)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We can clearly see that the model predicted some of the labels incorrectly. On the other hand, it seems that there are quite a few examples with no clear class, which might be either mislabeled or require a new class altogether. In particular, `joy` seems to be mislabeled several times. With this information we can refine the dataset, which often can lead to as big a performance gain (or more) as having more data or larger models!" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "When looking at the samples with the lowest losses, we observe that the model seems to be most confident when predicting the `sadness` class. Deep learning models are exceptionally good at finding and exploiting shortcuts to get to a prediction. For this reason, it is also worth investing time into looking at the examples that the model is most confident about, so that we can be confident that the model does not improperly exploit certain features of the text. So, let's also look at the predictions with the smallest loss:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
textlabelpredicted_labelloss
21i feel try to tell me im ungrateful tell me im...sadnesssadness0.017331
244im kinda relieve but at the same time i feel d...sadnesssadness0.017392
133i and feel quite ungrateful for it but i m loo...sadnesssadness0.017400
392i remember feeling disheartened one day when w...sadnesssadness0.017461
1310i feel like an ungrateful assholesadnesssadness0.017485
189i leave the meeting feeling more than a little...sadnesssadness0.017670
1120i am feeling a little disheartenedsadnesssadness0.017685
783i feel like i deserve to be broke with how fri...sadnesssadness0.017888
1368i started this blog with pure intentions i mus...sadnesssadness0.017899
1466i feel so ungrateful to be wishing this pregna...sadnesssadness0.017913
\n", - "
" - ], - "text/plain": [ - " text label \\\n", - "21 i feel try to tell me im ungrateful tell me im... sadness \n", - "244 im kinda relieve but at the same time i feel d... sadness \n", - "133 i and feel quite ungrateful for it but i m loo... sadness \n", - "392 i remember feeling disheartened one day when w... sadness \n", - "1310 i feel like an ungrateful asshole sadness \n", - "189 i leave the meeting feeling more than a little... sadness \n", - "1120 i am feeling a little disheartened sadness \n", - "783 i feel like i deserve to be broke with how fri... sadness \n", - "1368 i started this blog with pure intentions i mus... sadness \n", - "1466 i feel so ungrateful to be wishing this pregna... sadness \n", - "\n", - " predicted_label loss \n", - "21 sadness 0.017331 \n", - "244 sadness 0.017392 \n", - "133 sadness 0.017400 \n", - "392 sadness 0.017461 \n", - "1310 sadness 0.017485 \n", - "189 sadness 0.017670 \n", - "1120 sadness 0.017685 \n", - "783 sadness 0.017888 \n", - "1368 sadness 0.017899 \n", - "1466 sadness 0.017913 " - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "#hide_output\n", - "df_test.sort_values(\"loss\", ascending=True).head(10)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We now know that the `joy` is sometimes mislabeled and that the model is most confident about predicting the label `sadness`. With this information we can make targeted improvements to our dataset, and also keep an eye on the class the model seems to be very confident about. \n", - "\n", - "The last step before serving the trained model is to save it for later usage. image:images/logo.png[hf,13,13] Transformers allows us to do this in a few steps, which we'll show you in the next section." - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "#### Saving and sharing the model" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "\n", - "The NLP community benefits greatly from sharing pretrained and fine-tuned models, and everybody can share their models with others via the Hugging Face Hub. Any community-generated model can be downloaded from the Hub just like we downloaded the DistilBERT model. With the `Trainer` API, saving and sharing a model is simple:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "'https://huggingface.co/lewtun/distilbert-base-uncased-finetuned-emotion/commit/352c4147e4754f73a0b41f7b175f4a907270c9c9'" - ] - }, - "execution_count": null, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "#hide_output\n", - "trainer.push_to_hub(commit_message=\"Training completed!\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We can also use the fine-tuned model to make predictions on new tweets. Since we've pushed our model to the Hub, we can now use it with the `pipeline()` function, just like we did in <>. First, let's load the pipeline:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "#hide_output\n", - "from transformers import pipeline\n", - "\n", - "# Change `transformersbook` to your Hub username\n", - "model_id = \"transformersbook/distilbert-base-uncased-finetuned-emotion\"\n", - "classifier = pipeline(\"text-classification\", model=model_id)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Then let's test the pipeline with a sample tweet:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "custom_tweet = \"I saw a movie today and it was really good.\"\n", - "preds = classifier(custom_tweet, return_all_scores=True)" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Finally, we can plot the probability for each class in a bar plot. Clearly, the model estimates that the most likely class is `joy`, which appears to be reasonable given the tweet:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "data": { - "application/pdf": "JVBERi0xLjQKJazcIKu6CjEgMCBvYmoKPDwgL1BhZ2VzIDIgMCBSIC9UeXBlIC9DYXRhbG9nID4+CmVuZG9iago4IDAgb2JqCjw8IC9FeHRHU3RhdGUgNCAwIFIgL0ZvbnQgMyAwIFIgL1BhdHRlcm4gNSAwIFIKL1Byb2NTZXQgWyAvUERGIC9UZXh0IC9JbWFnZUIgL0ltYWdlQyAvSW1hZ2VJIF0gL1NoYWRpbmcgNiAwIFIKL1hPYmplY3QgNyAwIFIgPj4KZW5kb2JqCjExIDAgb2JqCjw8IC9Bbm5vdHMgMTAgMCBSIC9Db250ZW50cyA5IDAgUgovR3JvdXAgPDwgL0NTIC9EZXZpY2VSR0IgL1MgL1RyYW5zcGFyZW5jeSAvVHlwZSAvR3JvdXAgPj4KL01lZGlhQm94IFsgMCAwIDM4Ny45MzQzNzUgMjY3LjA1ODc1IF0gL1BhcmVudCAyIDAgUiAvUmVzb3VyY2VzIDggMCBSCi9UeXBlIC9QYWdlID4+CmVuZG9iago5IDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggMTIgMCBSID4+CnN0cmVhbQp4nL1WTW8bRwzd8/yKQYAC8aFjkvN9jNPEaG9pBfSSixLLig3ZbuUkRv5938wmXq4iCT0JxkI7z8N5JJd8HLa35vwV2/WjJXuL58myvbTnv62+3nxc/Xl5YT8+GgJ+Z3zJrvrgc8Ryo5eSsqNY8LbBXr36ZMy9wemwuMTBa2NCfLaKLuKnHUwuz7DNDAveETdQ2T5jYLg2/9qfjvU+uGKFswvBblf2b3sPJwgrzz5LDNU2hv5GYrdw0Rz5rzGJHUeJMU+OM3nH3udQJsc1VpwIiU+AlbVCu/Pv7CncZ/YuSaWQlP8xusTwJSv/FebFpUSZQ8OVvcZPGUJKjilGr2pHqDgqKUuaQtBYcXBJsm8RKHMFnzAA4eJiwccPKoBELiYvrIpfY8nl5Jlqg5W5gk8ZQObW2pG96l4UBvmaKajuVVhyJcHn5r+yntATuu/FO3x+wnJyP+GNE1ev3FdYdpmiLy3R2lzBzwGcv5JRR9eQOmgpnH/qwgfEFHEMz6KiVpA46szmAmr5ZC4W9vwtWxa7uDZwxsdApVap0VYXamriurgyL4fHYTlcDffDCm+PZ3Zxa98sTCc2jOwhhBRYNbzCjlFyE6kQsC/EnMEZY/U/OG+Hh+HbDhdmAdWky0JBR5lycYxGLBRKrDvRbcD0dVjNucTDHfS2SqSCjnEJvlqEslEmRgZgw/KdaYkcrpHF7Q5XxXmhtj6byBR2lK34XiMo+CB+J7JrcC132XwIrs2IWlR5KuwYm0dsqXAJOYL0pyr5MmyHf/Dc4H0nn3sGchDI43we7+ccMyhoa68KhA4RpIqhDQ/DjGNCDyQS/c8tlhQctqXKRaSOVHKYjBkMMUR8J82m4AN0kCYExYTrDUc0QeuBkS0cYcPYLDVI8nO2CT7IlsaNjC6PQhiwJY106QhdrQ7ySJzndBN8kC5WSi2+GlyuocaQEeZIWA4TCkgy1DTLvD4m+FDPjfGJh6L4kKkyQ807HQ80Ec6sCAL6K0M3OTnJpRVZxQUqSa+DxV2/kfYzXkMklk38BtsL/GH4gPUHFPoGz+fhG/D3L4dfhvdnP5jE/jHedbs+z2+6+5phz6XT/LXn0nq3/9KKvf/zzjvf+f2Aw6dSj2OcNdwnzfo5i+iUrq4kDkOqf3AJtdsibcjbi+F3ZKaNjyf8LvHcdbG9gTBZ5O0BY2XZs9ek8Qq/LZsWu5c919suYBv8tT1r7G8WbnhxZsYsvzP/AYOxdjIKZW5kc3RyZWFtCmVuZG9iagoxMiAwIG9iago5MjcKZW5kb2JqCjEwIDAgb2JqClsgXQplbmRvYmoKMTcgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAyMSAwIFIgL0xlbmd0aDEgMjAgMCBSID4+CnN0cmVhbQp4nMS9CWAbxdU4Pmtbkm2FcARTiIm9CRAcojhxQkhCuHzIjsGxjY+ElHMtrWwRWRI67DjcVymEs0DLUe6jlBYKpQXaUsrZgyMBQltooVBKW3rR+9tQpR//N29mdmZXK9uh/X7/0lir3dmZN2/e/d6MegYGekiQnEfKyZLW9nbd6lz0e0L+cRYh2dUdvT39jTuvfoFow48TMnpvR/9gS9UPa88n2o9PI2SPp3r6Fy81Z2VThGg6IeS0yKiR3njYRbcS8uf9CfEdM2Ia0Y/e+3wNPHsC/h02Ajf811U8Bdf/gH8HjozmNrXVddxNtN/A+4F3R41NaW1HLzzS9oU//qQxalYtOewQ+H4eIX+6MZ3K5j5+nCyF10+C5zrRaFPS/Pg1v3nv1N2P+CcpL/8A7pAXH77vLfH5cfLjhvJ/VewNX8tJGWH/g/fKF3zcTEjFPvA8Xf4v7En5n7aDkF78Syq0+7Sr4d2KssvKToDvx7FP7RSyVFtGSFnQV15eVuYvK4dnAbWP1p61a7WniF6o4GPtTd6AaV5Ln5VpZd/EGcyAv/XaZnv8cxh0+Hc2fGPXZdDzDfy6nBxAbuPXFWQv8gy/9pEjyOv82k/21fbg1wES1hr4dSXZXRvl11WkSTuLXwfJftoD/HqGcr1b11bt+/x6JpnbfR2/3kMZdy+Es5xoFVXwbWf3N/m1Rg7teZhfl5GZPb/m1+Wkvefv/LqCLOldxa99ZFNvnF/7lfsBclnvL/l1Jak7fiO/riJDx1/Kr4Nk6fF/4NczlOvdbvxi3xx+PZO0/OFb/HoPcuAfCs98eemSJSsXNS3Ve418Qm8xMkkzqxvJqN46kolnc3EjqfdHRsaNTG5zSM8no2ZGNzdFEvlsfMzUE/GImcyaei5lN4qk9M5kpHFxa2p01MxE4kZCH5hIm416cyKhZ+LDI7msnjGzZmbMjDZ25I1MFIcwklm9NQXD9pnD+YSR4R9NjUuWNK2mRLRatKVNactFvElxH+vMTDaeSurs5d5+HT7wciSVi6SSY/RB4/IVq0eNjWYqF2tMxIeWwp2lTStXLZ9ylDhgByYwDLgxM2ZUz2WMqDlqZDbqqZjumvRiF1JcjxWMhzzQjavQYmZyMJU1RtYcMkdyufThixdH7G5yFLPw1dWx3p5K5vT+VCwHHZl6GPoZBIzrXXy5moczpjlqQhu9tyvc3B/W+8LNbfrAms5+va2ndXBtuHtAb23uC7cPdnVtQDjGTZg0DgxfciNGTp9I5fWNppkGdERS6Qk9lsrosXwmNwIDZcwYICcZgWXX9YERwNkkMDTAK/o8+/s8fV54sKs5BJ+8Lb1UHsM484p6mbeArUzCHAYkGHbnQ2Zu3DSTCC2dh1yREC5JSI8uHlpsuFcupBuL2YP+XKPekolHTYFZ4IHMhN4QSSUSZiQHXJCYCOnzXO8DOHS0IYoyyk/6UDwZjSeHAVVJoJjIZIA1MoxJBA2nxswM0DbFExDdaJaSWp6y3Yh7qQ1cHVOPmtn4cJK2E22yujk6lIrGTWyQMePJojk4upq3IARsPWzieo7HcyPQ94Q+agIThPR0Jp7MQU+jBsADcwb6NWlPmVQyHtGjqUieAm7kgAtBaKSj0AyaGNHoolSSXmRy46nMxhBQ1ZBORQGsJhM6dIwUDim6ZsQ2akwAxnQjm00BjunQCBKdXDqTiuYBocnUOCUNmBe9G8vn8hmzCJViQcxYjE1cbzAW6OMjfCHmNbe2hnsH9K7O1nA3sEVzR184TLkhhGTXMARUFmNLRpEdOSMf50g3IhEznfNYEWBepK40XfS2eDZCO2qLD8dzMLd1QFcpenej3tDaFtLb1rUB2ulIEQWqVNpkkwLY0yA7szB9Bd2xeMKk0x4fiUdGPCAABFCqM+JJkLh6J4M/mgKE5QCL2REqvQE3JsUdXOYcGAuJlul8JjICUiiEU81m6YPxZCJl0EWLLkbMZ2FOCQo+LiH0bXqTKUiFpkZ9MA24SRsTuDAAfyyfSITcjAgQQpfDGSPJJI4BwCQXSQ1E+SGeNIYABV1SGRUjAakIIKcLkYybWTZNKmgaqVxSOcWBuUR8NE7JDTrtBbWSSho4wU6KL3rdAmCA/M6yTpKJiUboI8MIM58BW83Msr6dSHXxmp4doZgboowbo+sElF6Sj51cWQw10kGIkU+e0sqwmTQzlAWZBAGsxowILCRcpQCz6RHKszgQMmGI0riRToNyp3il9EhtUDq0WFrRvHEaSGFTGzVBs5VENNUdsLoZPZLP5lKgTSf0tNJxXHQ8JDq2cYvwAjlQ5gOcUTkhiS9NiSNHl49JvBAHhtIzAoR9AylFTcFfgvaioIwz8aE8FWG0WZ6B7gA75LFEsRTHE5dxxVKXPgMpRykxlaSkjw/g9Zh8IqDAUVGm0QcKa/RyoUffyRoJuuhGLpsyQVFMjwBBao7Eh5C0neNlBUZC+tCEPk7xGQMEG6NpGIUiDjmCi/aBRdmReCaXpZiIZ6KL0jDrCT1r40dVCiZK+ZA+mh+mKiAZH+WdmLlII5OhWSHnhR0B45lUiyZS47bkovSST2bzSOmMEZB0KK5gjiirxLqHmOQDRBWZoxvoOzEjnqAdAXeLhnCZNc2NrI9sli8/RWfCiKCRZ+jZtBmJx2BSAou0FSoeh7KRxDZkMhoG6U0nlKIMBG1gyoBhEGMZekuMSGUZJze66hvBckBwkYhhTdC4YQKtmJPYmAJ/qDAoJ6B0iefANQXbjOoQoZ4ZcXCRwMwHb6mgSIFIxkQVDLDgTDIwg6gey6RGvUT9QAkYUdghEgDlggvphPPpYWpTZxXLIWMmcESq6uCr5OtR4C1YCDYVJriSE2oDME3jaHvADEQvDvsEdNFSB20ZgqOBIrL5RM5IRiao7QKrlItnqQ1ngECN4Ny5UqQ3ce2FPxTjEoNhCIgDbHcjEaO4zuYj1JSiGjeem2AiBEah3GDq1LjOUys5O5LKJ4A8EtmUrXmpdcv7B+mcScHK0ndijDzjTEyCgxYV1FO0GIgg1rZUE6bzNoJByJZlLJUYM5lccHAWdwWdFM8WWwCJStuWvQinLdTp8NEU79TclEOvIsXGAaTlTBCOGaoG4CKRmjDNLJODUTNNpwgAC0M6lcnqeb4CRdZwVpktFS2MAuhIgL+xOFrkcBUzkdEBLtsYpc6Mre0cTSiMYDPg9HExI/EM0FOWEgoqJAFzRhhG3kuMANL1zZhoRiICpr3W+SxQIdDusmnaG8W6QxgdCAPVhkDkoylACJsz9JBEQc2gBbs/h1aWh52vCoSSKp5Zp3EKXQwdZzoXh5bldhuzVhIwnSFAgd5AZwcyGcx2QPdQ1syJ/p0NwW4GImtPGNSgpYaj0EJyECZ5zU1c8eELw8YoJwKKE1z3rATZzDDDACiU0hoqA6/J8RUQ4rJI53O1xrsxo6ttkwIgSKWiWanIjSw61PZqCfcGegEzZ8RIDjMyAC1kgrpFqjBNaZlRqQVeRYYLtzWpcWoU0HW0l0tFSBzlXNJkk2N+njo8AJzM8qk7fAzHOHRd80khX+lyoZSTQzbEEPNoRUj3URGqFEDbukuYuRzTiTTUEjHjaWbp2NxBFwEwgaqS4Y8NCLopkULKZW4U3FDtVemvunAQiVOZQqebEqxkD8T6p8vDZUPIqdgFIlCbF7MZ8OihYN4s0MMOI3XIVE0wYZ4KHSTFLPXNbI6DUScUyGG8+QcsbVqyekojnDZrWq0sB1oP8ayDLATfG0nPiVDIbe9kKDVmLtAb4gt03hPaqHqW6tSILswHkHhArc61FzAPx2MMqgW2hUv5xoxGnd6tAI+p8IZ4fIEUJS5CNiiRjKJZZAhVlUWnDBd1LG6Oo/qGfpAC6ZcG+o2uJLI5CDlGiwlKJ/TKTAI+IngJYKK5MUHJy1blnlOmExZzEROYvq9EwVENopIydUi4SVnu32ZM7tvA4BlUBGgax0cxbgMUA2ZF1hxJJaIh9sTI50ZSmfhminOhaXGm7qfU5sllWRQX5233JEYLKboarXl8Qxp7ABiYakxLIJ1nBbexmBZ9B3gpGpdm6YgaJqTPucZiBnCK9jpE43G0y3iOLipTFqWMm0Z9aMEk3n5Sz3a29+kN/cBoqOs6k6NAldiqz0yD/Y+QYKTR5lBFxKKeZ7235oG6m5Y3rUbv4RjqFS5CC9uW05RUcZY0BsaIhdIjPKc36AvUUhdjAo4iI8lUIjWM1mwD4yE3ZPqA0srmLsnW6WjMFonoIJubMJ6UmPBwlU0DVpnbJkVCUZon3FoHMbe8sTiKShkQyIUGjKihxrw8VJIhZXDwigzKvyJDwPyQeI7FtQxmG3rEdIoCIcXhkmJdXBzUtEP4WRYvcepEfI+2bSwW4dPDJFI70iwKmhSLl3taEcySYj6VgfFJGkQAm9iggOXTDB6veXITxEXvpd1Gjy5E8NhlRNs2kIi5UXvUpI5+ccaD98HHyCcVCaIw5TTBx4g+NTtiVBYzCi8e8uCsAhpPMKEsMZhbQIdCHxHeBvsWSA3cihEjM+pISTE/RKyhy3WhsIicFV2/omBCMUgKIBnq5MYxsMIQOcYn4p6bLSpBYeQTbPnhnaTBQtm6fphKglQIeVsQlDhFMNZgeZdSgWmKIeAyDCHj/Jjbw4KZ8nkiFaFZFSBOHRCIca2GeV3N3fMWMPWKsUY6BBeuqCGK1lLil4W6omp0lb6dzFNNgnSRyjFXK4NaHqw6Sv/sFZQ73g4oCJxGs5FJYtaF7JJ1Nj5C2Y/61p6R6ZBtC5R+fRRF1IgxZvJQuLd0ovqYZjOS0Fd8lK7fiv9IgtAXKHBom6CsKClFJvVFeDARNFuSCkElbthnMvsp5+1NY3ZAKi94z45z0LAmPqCCwcD3sxNZ8OKzznWXBOsK/zNuMKJM+wPiqU+D2kK6OeMZipSkktxzCx9AEYY4wKHFRtEQ9/O96B8VDhphOVN46ioAtDGfZ8kwIvfEzsinRDCpmVuO6H3QZKnApVTxQ1QpgiZFVxCTvpFsyViMN+6KY2qCOoZMpsOolmG6RHWzmT3p5cJPkj/iYWka2xWx6hCCJixcNL05PaUynka407IW4TpmXONaZqV9LcJ5GAgawbxzPB3nVIKSRZrUGxTG2JhMUfs+McGcVhi35Gw9tQ2gTMxzgrs1GFBjoSc1LDZvxIhsDM3DWfEp2ZMJOfNfzGPwxioohtSYF2liKJXCIieqr5STpXULKFcamhaAZRDZuCifZtLAnaFg9AlmG1gTCSVAlKT5Xspe0jTJpYQZJrUpZmOiE8I0ByyC7zJmMnpgsS8YlbMyz8SpQRJFLqMEg+6oXIJ+GEbAMhiOU2YTqV/8BK3ForYlbPgBt2UBepM5awnGbzRenGTkRkMqckU8DBKguXhWMU0c6MSKAjsRPZRBk7gousbqI2iaEjg1n8gJPS+9riLUwOs8WwkruwrDz8mULUKxnbkJmDYrRX+OLjp46zRzjOCLBS8mHxQATDJRYo5HaAGL4kjDE+cLDk5OenKMcPOz+CXljlFhdkMJkuhDYLIYeTUQznL1kzUESZfGlWQKTAzNRbhiV5buxk5ucI84LqwqQXbTpDJmatP0ScIUy0blGDV9GvXuVI46BDTky+hNYSMBI5dBPINEPaZ5CPZiAJBGIubpvW3tivxlydlYPJPN8QgFraGhUaVRcFdSUQSTyzWmYFVJmGVr4sIG4hRRRTPuSxjrFBkGJcz+knddLpNSk2G7Sp/IaaBhbQYcRhNY9FXmodnYGewJvGEeb2ahyPSE9BsZF9FZY3FLnsVQ2dgJY9xe78Ekmp79OZZi8uyLOyriNdZ1Uqgvg0Ue5IAZ9gjDpqB74zRR0IwZ3TQrOeBlU8hzHuYerCHKdhDU2bzBGM0ZBEERyqs58FEWjAwJdEguTEhBBksgMM+HhmEiAhjArmlggkpWzhRVajQxzeP0YNG2CKEzn81xL51aUNTr4sl0RAEjfqdeVwwth+2cLZJMIR4HQyYD+0UmoBFbIT4IDepJy23cRN9rtedDTFhSnobbKGKxKEsYY/CSHeKb6jEaiNm0ERExxY2w/nC9mtnmSQxt83aMPVWDM8orgJixFOL90jyjkWbLmMwmEMUZapxT/yAJutKklkaUFjGl41SzgWUHSwCsjKFVZog47A+w5s3RNK4Z9aNQeyPdpPKZCKWcaKl8DbBxCjDOJFJqiFIcnY6wwpX8OJorRT5xp+rtSmOQvQLibhikAiZ9gCNCDrnspBa8a5OL8JFpGZcSEgRTyuaIFJcV05c9qnuVNWkVErUknIh02JTTo+ci0xKLJJiB2aj3IGuengcPIBqPsJcpEMJjiTn9ERZQYFztqs8R4LsphRE7pxXPCAAzZ3mGy16nbD4Gk4szt4ARKF8z9D/HjDgLj8aciUeKXclySSYR0SGkje2sGsxvOGOM6g2UnxEEzH1RbSY6Qsn1iYCzbYWhiaKVRpUuyJcGNsEkWMCCQIo6s4MwIhRIi8YyUWra84ghVvo5Kn/TmByMqOJKx4ySTSSCx5Rq4XFQcwcD1AaNDShBaW/VWqLWK54ETZ9FmpfoGjKTZizuFSujQaSmpfZwk5cNi0rKVCKBjlVRopu5zIwQGf3CKnT3DCgarZQUoR2riW0W1aFPZDyZJs3mH7BqdXNX75rmlvCA3tOnd4UHBsJ97T19axf1hbuaB8Jtem9fT9tg60A/tqX/4KneF+5v7go739Bb+8LNA5093fIVeN4WXtfZGu5fTVMD9O3w2pZwW1tnd4fe2a2Hu8KtA3093Z2tsl1zV5fe0byWthjsD/frzd1ti9V+Ih799Pe0D6xv7guv1qPsaf9ATx99xl/uD/eto18H1oT19p7uAfsFnU6G3u0d7OvtocP1tOtrm4/zbiww0LyuubOruQUQsK6zGdt1dgMSugGHg/30VXytL9zb1dyK5a36QLh1TXdPV09HZ7ifoj6+QO/CJCzlMCpHmhPpEWPIzIlsoFomlpeZJ1U3U9OJJsM4oWQjGSM9lEptxFVmhWa0uoR+pWY7ditSP7RaGhiZGm8JGw5aEZS16VK1L8RQMAyL2Kk5fVCh+SEaxQPKHU3TwtW4uSiSzykNsiASIvGEcgeswlwKJhGjBQUslxwdMVGb0IDNRuoEC5SoxQOygFkk7rm1xoTihEgQxjxLpybLX3iZZ5g/wrAN7ZoGKDCVyN+3AYwYtALJxjKNIOk0kRqWaSe1wBTjOQ5nRcR0nImvIjfIkcbiHrcz25YVoRcQCgma9xQRJTssyZwdNikRPcpNpwRDTS14hT1LxdKaRSyDgsrzXWJIbO9l8pQwJLD8NGPbEtPOnuncaUTbhdppGdvrLJbgcbp0DusKQ0WA9iFDrdGl8TqWZUS1S796F4UpaySqVgFbci8HDR5GEFciAiVsfLuQP5oxqIqwtYKExl3gglU3idRwii9ViIWdgPeMyEZjmBMPtEmn4jzkZyR4BRraBEYsZ2amRikgqpO6NZgkSHHVU9xMiUyVivEazNI4pnjbDU5qNIW60jZFplWoWircKypLsdJj8mK7SWpb9aZljZy54eWwDOe0sSJcIbuzasraIwwp5FlyEaVI4AAgzXTCEEWNrlAR9k0Zm5bDOHNHDC/tjjoQXnchi4zVwmJkyOLqYVkWoW+Mp7JAeLx+y64uBr7PpGg5zgSLDqHhMmpERlDYDFNvmc6Akxl8H0o43o9ngN7TfG8ISle2csidI8wOEnlazqHOMBIX4ZOL7bijiFZZgKwadGfZT6EvlM5pn/FYDGYfjzkrCowcmniHFqUkvao8eSJNidHZpc1w25X6mmybRJIxAdYY0Lj0sGlXEIvVp2AOTWCdIaKKJxspKk1ayJnQW3sHbeIRprDY7IR8IzhXVUJqcNBkJQtYrkgjXynqNPCq9ow640Z7q4goNbZLsrhEiXr6tJNIAspzy6eb3hMgU8SrGT7ZxBFrySCzJ0zcfJPFYkfcFYceNHNV0UARAlvJ9pSoCMB4MvVnY2jGJBh+ecraS9YAIXKjBsuNsL4J65ddwVlWEGbanZsmd5dESVCGVge5KncoAHK1vcqAnGLPrgPDfUwirjt5PFf1zJSm2VwqI3l01AQynmDaZMTIRFn1KEoGur6HFTMVtKCVAlm9RDU+ZgXS3Iqm1atgKlGSTEx4eZbU1HHUpiOFgOVoAqIbVi1ZoEeNiazikEXNBN084J0GaNQHeDjREHBO6JGEER9VdJ4Z5zo+l894CSruUdpcjGzBjDPcmTbEtpXROLurrsbOTyBtUR2etWs5UxlWqh/hhc+gYKgxKDUaWojspqAw3m8J9z9jDtP9uYATVrhGe87GS1uLtNQaZQgP+ibBQR11eOpyh2AUs6jutI17IynbM+2iD156Ez6hFxzS/q4Neltnf2tXc+fafnQjwWnra+4eAJ8rJNpQj1DvXNvb1Rlug467W7sGqRMZ0lsGB9C37upc20nd3oGeEPPsWFulL3QRw32ta+Brc0tnV+fABuy1vXOgm45AXcpmvbe5b6CzdbCruU/4lkDePWvpi53NXfrAht6w3tYDvdFBeecwYPMAjtrTG+5jvjQMVuyIru+E2bWE9cFudDz7BnspyDAuXPf0LWrvC4dDOnq2vD/n65390MPAmh6Yclu4HVxw8EkHu9vCfQCM3trZ1zq4tn+guRvcbb1/DUWkG3AYuqsTHeCBHn1DzyAdq7l7g94DY/Xh1DeE9PVrwvgV/PNWGL6vuRVDDOCZD+gNNub17nBHV2dHGEZbQB9jF+s7+2EC7bzX/t4wHTpEu+kPHz8IDjV+7aF9t3a2wXcArK15bXMHXWnZdVdP/wANRsDKwP3+ZhoDwKBEC3XT6VrZ6KOobgbqoJGNwa4BgfbB/rD3CoTXhbv1zna6ep3tlD5gls1t6yjS6Av9g61rdFj0/k5GIKUqxZCbqKjH3VwioMpksFKllIgbQ/EE1Qd2wdOEjKS5qnxssx1czIjkLCeD2nzGNlokU67oE+vczYQHo/tTXI0l4WPxZtwxSGte47ZxlzFj+aTqC0dSWW+HybYeMIUviy892/IRbD/dRopRZLcfnHX7f1TrrGjkVNvVvF5fhCQ2CLQOZNLdBcvWT+PgziAy37BrUC2hJuGY3pZWNouziNJ3FL5ZGhVlRhO1nqktSPdCGtQGpcYxmO7UNOTqJM535RTfZRtv1P02TAGwfXZZ0HDUoKS84xJfG2wW0/s6O9YMsJgdPG/ZgAjoZPPWw+1ULHSukwEykDRheB3VZFvzBtkhSKrOnrZG2gj4gvIGCBFVWvZRpLZ1UhZD2UnBOq4TBGZzOzDfNDt3BKrVnXxDmHUsShFTKvBCXMY2+4ZUHWQ7mmhj0zwCNQmWcZMAbX5eIOI2Irll4rExccjkJwvITKqaH8XEKP3SbY5Dz9TLH2zsb2xuDNnLS2O/E2LzJI3OZNlGcjSpMOdBIzkp7stSba0EZFjVBAsPc58Lt5zlcLt6DKxcZCkECW2ArEi2CJVdsklMCcOpvOHGEFhkaAJwuSTK89VXbMcsx8vE9JgZxW1HERCNObuo0wtb/c0hWVyD+zxZpMaMUfFBF5vtJ6QYEMOgz8vHdcKR0cfMZJ7DwE19u3eR+Ie3ABFYlEi9LSOecEj2EsRhqlF8V368m4f2W2lGMymjUmLBRRKo05H/7qcxG2jXgbEfR+wNpVEUY5BNq3hqOUJrtxJmdFiqHVZgCU4yK1BArwQjjcWuCXc+0JMWyQXbkXZUODk3DGQ9XQ4GkgvDvJaEV2nwuhB3mhFpRFUFxceQlMo+cuWQpzRoRjG7gr5VOpVlspjtbTCkF5TKsAc8OBlyHZ9BXdR80k7MoCBG3DBaB8o/nTIcrfbkFbeunZzdKX3MoMkZNXsl/HHXEvA4UTSOW/twtwJtiN4PFWPUUI+YmAbk7GIMZenxMLzu0x1oxcJJu1YJ9RLW9koXzm5qv8q3UXPEo+CkwsYbYozRcRQzqeExK7rByASHKcPko8E4nkV8RtMmSnoHizICiyfHjEQcXPKxVBxzyPmkgoAQNzZGeUrAKbY85TQVrnHK+fyMDB07Y7SH8wzZxTp8bGVatqjnlopnZpR7bPItpHfWeZZ3nFPoumgYuol6SbF7rO7wwIAP64zV11D9MMrT7LxKkrJk0ck+WVFJTauoHVqD8QTbVyRijPrSJlbm5MaikacHS+R4aFVU5NkEyQLp6NMyYnYWCmZ49R/nTPo6J00mNXS6LxXWBfDI1FtRdsR+r0QhYPHJJx7YU6skGJ977TPgp3SBC5zDKCjoRzdOMFFBo+lDvEQ1Lup4KWOJbROc8JMqMlLU/2acVhx3XrpUrRVi/EB1O90iKSoGUHjGlXw6zIVq8BB9atLwJlZe0WN66MY6t+Sy61DpqRTpVEamlOyIFsWs2Ag14QjHu/frC2KjcDFbCLGTz7KiUYZIQ30kLJ2QU7HEjFHwLwyKIvSZQI/HxAajKI/YsNWT9WyZfMIUKzeMnpSXpPZOcvBNbll3tZ7BCt54foYm1mBVaIVDPEljLXGxN48anMOcvWIpEQBSTlDg8Uss9xyLG7q5iBoT7EyAEhFQSmFFksUcysZzLN5MY0i03DCnj4+PN3pkU/SxJauWrGxaXvKYs8V8+uSr5MtkKVkC/60ki0gTXOuklxgkTxJw1QJXGZIkJsnCNwOuovDZSkbgbhzu5eAvvauTfhKBu+PYPkc2kxDcy2N7E+7o8HcTtEjAvSy8MwbfdfgWh3smtMri9xxJefQUwbud0CpCGsliGD1FRuE/2m8Ex6eQDpAJkoZ7jXDdDHfoPQrjMPSVQ+gzOAv6Fh09Ci07ABo6SlSZhYHQ6DgKm20ftB5GfNC2zm9N0AvFXRNZDW/0kLXw3+qifkWvos9Frl52tf06nAXFYwqhVqHohbd1/k3eHYGWOcRkEmYv3mgky8kKeDoKvW6EPmmbGNyl6zIElMDaLIXWK8kqaFuMrzinC4bdYU4TJn6L4opm4CmlgVF8cyPcS8EY+hSruHgKKpj8bW/6DU2TciWdt2D/OY7lNXCf0s8Q/KM0lYOxDgdIFyNsbmhyNj2yp5NDrJN2HCWH8FAM5ThE9FmYwzPI6VcnXS7OaQbc07YmjsD6oXzcBe82Q49hpGN63YajrgFMUjppA5pthX7XwrNuuE8x1Iwt2+FuF/y3QcHHOI5Fx5EzTvJ1HoFWdNQJeJKHz40ITZpTB22fhmc6zCyFM4hBqwy+x2ZEe41xykni3BpxDhTaEU5nnwwPDXwUnczzeD4P74dxts1AI+y7s19x1/ttNh/3O16wzCMLHDyTQJ5hlGB4QE4pLYdYN5ECBW7FenjxSEjhEnodBfocgn/GlDwXwp4XO97oh34bkRMo10RxXk6aZTI+g2vbgBAkcF4R5DAqaRPwLIQYmnx8hh0xtyGbykwbX0PwThKlTxKwpXN5xmRM5BNjrNFBY14UNAytxhBqphsEPTFJN4r3mFTL25psZEquNhTeMXGlTJTpwwg568/dTxY16SjMM4V4MJUeaK9xfHeqdSgNFV2DENfEw4hLwZ/j0Av9xuCmqz2Ko9NVpG+kkUKSiJMoPjU4ftg6M/lr2jBlEAbKMXTmKfjM2xg3EGb6nFkRaWjBestyKo2iVkzherA7GZxFCnVMiMuqIfgUGp/xpmrFiHmklFm6oVYlG302wWlMR12QRbjj/J2oA0ti5dI40yjMglFoEr6N21KDrZdoS2ViDuWiOQ2qdHMIlZ8xx4pTSjCQq8ZxBJUjqDRrhf/CoCUGUJJ24rduri2aQdtTPRC2dUNIkXYNMDqTZTEHlwnKjpAz4E7cRen0fgS1Qm6aPMI0r5RdaZvT2xAzERuiNuScOFIPXbd1XF6l7LYbEe5W+BbCe+vgH6N2MadICVylUD6pKxXh0DCbMstX35u6Y/CZwLfYao/jqkYcVDIZDhgFCFlnYC/MfqWSXsV/FGebRIpmtJjlo6QQuiSnbZ1zaopDUIrGQkV9ppE6I8gTWVtnRDhvZe03xuGNBHwaNqdRrSJpPsvXKWFjX3Ihg9vcJWnKbIUm/BwEKBndpJFjJccw/MfQNkxwe3Byq2wc1y6BOiCD81BtHINjJgmyyMu/EfqBjmsAxzAqcNsJch2mogQpixjOBUckURMIzSRsJWbRMIxkOTd56ZTSNEf9gFHkKaFnKKS9hHkfKZyVXMFOm77E/RaOjSSnDhWSJOqjRg5HxiExKY3RNczyWUm4J6PUyfWajrwgaG7I1rgxm5+YTN91fTyZrpwOrqU8CDmkT96WK8N4x0QKzCkrzWiC0WoMacLk2nsC8cdoNs37d85IasKQLcdp2zT3ygW9CvlI11RgO+vBte7eG/9LlKKuGh2f+Wy7TtHC72C8m0GJmkc5lOK+6QTSnjfE8SKIh4ogLqZbiV8mHYTmY3Qm7AkvyZe2JUfO5j7Vxgu5MCPks8SQhJtJpahDknp5rDpac1nUXnGcn7DCRG95B9ZLYzs0TS6KKVzERnDacdOxdcV7zJYTMjGF65JU7IwM2gIxhRKc77hxIecq7TTxhrfW6HVZemKcLD4XnG7gfFPwnXkU/00JyGzNEVw9KbUnm1+2iEZCSNsTCK2gzxinYANxy+YiKE7qCKfVPgA6MYuwZHDGIRv+DNruab7WE4ihYvop5SmYii0fQorPAy6EF5BEaJyQUMkcQTxLO1TlvyEibG81HsHmZ9q+aAKtmmKbS8iXPEq9vCLTVY0gpY6gK7aO0q5y83uIqDYfo6ipI58b7HFiaCkmbIiY7nb3yO5mkZI2OuDIKvFFJ3Um8G0ZyTOwhzSuVhz1acSDFkVf0uMp7dl4SbYhIixoJoeZ7S1WKGVrIJP7pQyuccTCGF8F1so9R2GXOaWb4PWNPOYgsSslMeMTGblRLbTp6CR1nm76kx6G0AnSdqH4HEVtzeJmwg9xe8+q5HBaCWr0YVdsBW9bIILjGwq/TDjWJMPXIIrzzODMpmvVD+wiHqVlJymBUblbF4oVpvEFNvuoLVWKYw4ZlAFyjsKry3LJW6yvR7neinF7Sq6KanEleVTGqwcWNY0TU5FvUQ9YSsdPmF+0dBK5ZRC3jo5wKPMYxTYQaxN23IXxUg6pUsThDG6hRpR1d3qKoqXke3fuR2hLwYUqDTHJweLudJ1jNl1nUduKqJTwceNI1aoVwuYidIPJW0bt8TKcilLYMsq5LYt04/Z5RezWCT+znTO4Yhk7JiTnxSCJO6xJlh2LErfsmZozJAWp/e5qL6qft5GwCKHKLWOERhHHFOk4uc5yZvwmk/EqZ7sxKT3tYrtX4rPYUhezj6J1pUJKbReBbzE/OZ8hPmuTW44Z2xswub2Tgm8mEZaYWMEoSl6xigzD7oh0CmHUbV9heJrronqizrUVVosqA8ScGP2NYeRLjMbuxYipaHSGr+LIqMjMFPt2pXsReGRxBrn6kjMjOAMmn7K2RJEekhvPwjaXEaNd4WKJQcG/9LuMRkYVbP23+TqPtv4wEfGoZf/l+MZ0/A53pEPiQfiGTJKP4hoOu9aZwZBULGoVtyzenyMyljW9eH4pC2HXvXg1dhq3cSfyhlFlXUr7ss54mxpbSfDVGeJUQGPCYu2Yncyi7Yy6hxD6XBH8k/W4gEvuJM4zgZTH4kQi4uj2hbxmotq8VLY5PT45wjCus1MSCDqR/F484rgtc+WsmAwVck16BtNdOScPuK3Lqf18p7fmhIa+sdoeQ2pBhoMU9pslXh65QURu0dtLdmdv4lxebeJyI4mrrcYUmFxl3q2UFaaCA5XfhK0V53LTabmtISnMYzGbS/BjMXeVopC4Ys8lielYOTWfV2r2Oa6Js65VL53HKD0fwa/UQnDbr4K7pC3nNcsGG8vOWIRX9tHbUhUYLI7dJZCTRXZEriWrdaA0yCJ30vpw6w7BCYwmpFep0p86Q+Y3JTg1MPjUbJTBOdQ7vuqVX52cDiJ8TZh/kCiSg6VmpMIvuMdpN5SKRQ55UoT0zaejzZgePZRHb2h2LjxJJHWIyIyrOwrmjp66/SAva1bkzYp1HJvrRAmcs/nNJwdg7dQSrMr6TyPhorcmlHRe3CFjD3FFexe3dOt7ZoNNd0UEzotzJ0OEVUksQGji+OmEScZRdW4J5riedUcfmI3HZOtkfO/G8zB6RSquFhDVZlctziGEXHhg07NkVC+czjFuZ4ydVsnkEtmwJckokdEio8irYjyXtP0yxqljqCPHFe+bwSNloHjSYD8TPCm1ObPkVLmYsOWJuEfnPMLtdXmXYVNGNyZs6VXslU9/lcUKu9fFvQL/F3klgZ1SEaJdt1OHirJJwp8T+VvWTs3bsJlnFI9ARo3jnC+jigfCohUUMnqV4JaffMdAuhtB3MTJZpvO3T6tXNOp3hVxnhz390SdrVzvYpjccwt5wqDG5uUYXpE9hjEWVVN9CSnPRU7Wu05LjMP0kuACd7SUvecVmxbvO30sNQKcsmEdIqI+TkAZ5zyewTGlZ7GrkZtG7HHBJ6JNJoOywC3tpA+lRD/XaNKvo5w0ymWl7KsP9S6L/0ucyJrGYh3qbcVKf16FvRW+M9ndRJaj7Ja5h2PsXOEiJYZdbE8LqSrXUtSBqZJFyEf2vmgRs+eZ8ZwnoyNK50mk7RRqG8FBDQ49NBXOaPTLu69i3eWlrWl0OkaKrUSZQaa6UtYnJRR5NHlW2URIR4g7bjK1pegVPXHG1pk1txw/p65FFRqQSRdRYSQiamouT3qSoRIzZ7kiw9a/7v0Baj6EjqvWa8kn06/TmboiZDrVJdPxi6dTqVlchS9iklP7iXI80S9dvamt8P8mTUrZLuWstGhEXjPtsJKmjkWoMSk1T8XsFRb1FpUILE5s2BjL2zVmu7aezijI5PL9k2QbpweFu/J48kh0cRzIXecm4qPM35Y1A1NV16lwOOeRRwvOywbx1pT/XezLGn0R7YjZdrEqw6czy4N5fKsYa87dSdIuMRzZAjErmUdMcouR1bjkiMhWjODV6CQ7oNR8iJsPJ8+6CLy491AJ/pu6MmE6WPLGSIaYtu0rK1ZUihxzrchU61ZsVTIPI8+jUmKF2DhJvCersul/h5WUgsIS2pUYhJCc7spYgQFh4+1axbSgIabLZBWyXD8126NWZnq9n0A/TOxVYZJT5xQo67UacB9MM+nmuzmk9yrrGsUsnJar9CGm5ksv+lWruqKkVO2qGDuJ6z1ERLwvxluI2JjIeUhfnsXqhPxXR5H2zq5kQJmF04g0pdrEKhReUKqQjaN0ZNpP5K2nXzMdIsVxgU8y+qhiRVG+GuOyXdpsu2I7Cf9Y7M1IcrjiiDnGfyv+f7RBxAgCczJuIu2KXbdFPnlexFmZyHy2JBGWoHe9IfVJ1PiTjBhN139k6+nleRlcervrOUS1pnxDWAyGMn4WZsF2kI4q0tR7zYol7OTV/6puMLgfKOxcg9dgmQ67xDubM446hVEK66s4SjCV5cOoSFZxsAyt7Cmq1HOyfP505b/0cGQkLKfMSuTUS2FA9Oxcz12vRnTmxKgfySSLmt9qdsUcZe5D7Cx106WXFz9ke4oRbgOJrKDc6RvxiO58Mj1Tiu6mU6fmlh1MSkk/TPgyql9SKputxienm4X/ZPuPnNXSom7XXVcdUrDmjuHKqLdTPqn1JlNHwieLWbur69TIteTLrGf82l2dJyuCGG5YDjDOK/FUWSJtFq8otfDf3Ku+kbA9gMK6kh46HX2Uxwh3dW2n79vkuKxxrueEA/dqhZpa9VSqWmwevhnBvZfzlLVyrlLxysj18NLVao5hV2iVeQwpexfWdHaHiKpUgRevFdXJSs+VFecnSHulgTSh/TvEsbIIIwaqbTDVHgpVfrJoG4tNiIoSdwURs7PV/JRX1IStljMa5uWbyr0xUR51UKPmjBZZ3kX4EQJvat0Xm6tTKzv3xJWqJPG2l6UNxqAT9hKDR6WRFPcl47Zmc+/6ld+Zr6XW2u5aHF5I/tIxiyzXS1JyJRz6TdQXy9wdy6kbRN2B5bSephMhYXIujmviFTUpTZ3yjILiHdFDXD+xVZ66dk3iWu6mZDqV1fO5/XmvXNfUVMNGd+6tZDy7Sql+ThbZrSn0H0R/lBeYps16Wv05m9NZbl3sOZbYd3P4dKSPtABUm0lI5jhyIzv5xTsjzd6ZbITSOpnNeno6xp3NzypPUmSqOiq5d8O7koTV5TDaypNSFeEyC/HJe2Q2XVrhSdUDc8/aaYV7xys/CTTFOzeyLhy7Y1VuaffflWVqVFvsPmF1UU5uE/aYiPpQ7upGnhIZAlHlq8o3b23kxqPTDnLuQRI5pnkKthdzDIqaCGp79JI2mKW3/avunKWWX4brEbWGQpxDI2qVRonJpWxUwabTXlM92FI2oYyDCL+yNG1IOpVUJfa4L3FonakjBrsW7d/1tpNnmbzPySjOKv2/yzSIam0Vc7I2Qa199doPrc47o8CU4xJUrUlQqyKpTvDKN6q6SKy1PLklr8hP57ypPhj34O9BlN8Cjn6SI+oupunD5cyouEdToVY1snOfktjDUzzDjOMtWW1q8rtiR0EzkXt0pUSTcsNdxTfd6B7jQ2m3M4s6i/hWNdpklSDSCnWezSHfohbcaAlMhzw5RsVe8f4LJg/UnI+ohokUYYbRrsmtBnUfpxPGqc/UaHL4PJPlYGXcImRzfpzjSM2lixiUyHWpXB9RqECV/JP5694RrdJxZ/nOVDusZD2Y1GQs/uK1A1rSVsg1E1Gp5xVzG0eJJPJeq3fhTbnDUuhp1lpasfKkLHdkjI1UXMX3n74tI4hZwk7rcdcpbuT8z+7Lun1xOqGo2nb2p2rPUhFOaY/LaKaILIVc8Ir9jAba/5Ibkyg3JBUz64LNmUmYJPcrTSJiGlFi2hZT3PbZWMyOcQHTyrJqVY2IlI5/GFxijRJxYlKKyHyU9L2lvEnx2hwhc1jceFf218S5rcNoXLWRUjADIePE6rhj4d77x2V0Zeo8cadDOqm5Xa/IoDoKs+6Gua0gd/owHSEkq5e9PJlskW2LpYs7jyxO4/KuEsxwOe/WESmXXfF/YfeUyl5lubWhxiQmo8jSccr/pnyeOmrJPP4Mt9lFBJPST4+iNU9HbqCwRxEqdWSBCXeOJVYCWqOonkLV1ZOfn+PG/lQyRZXsTrky/RoANTqrWtQ5D37KYuUlWzkZ42b1E1KCOvlM5j/HCDtpQlrIzr2R7niloF0vLScymkw6ygyh6Ll4rxpbP1a9QiVMg62fJRbkvi/hm7khkjbX/zvMFccVmL08FU9LL90tfUXFJosSLHBUAnl7Z8WVMO6qQHHSWAZ1ZZLrFtVrk2f6lT7zN03kzkER9y+2rnQi9ygVSxK3HvM+W3ice3MHc1wbdt2Ad6X0rnitu3auF51TnvsiUs57UdcQYTtGY1ymTke2i0qkJn66g3N2/8lpw079FCPsJFGZsZp6R7eaZVYloip/GS90g7weIN4+2q7aIgLiUju21Vod8Y5XfTLtgeZuaCX2KrBPm0kX6SVr4LOFhBHaHqyb78JvA/C3D/BJ763Fk7rDWAFF77cRGo3qg2dtsD6tcK9f6Vd8tvP+6Jv9OFp40jHo6rKznAfAWuoBHHqNwt5vg3br+Mme/WhrDznmRk/3bEFI26BVN+kgOn7qOIsw9taHY9A+vPprxlOidXizGfoSfQziKaL9+Lwb3lpcEp7INOHphx7aAZr1/HxqtgtOfbcfntJRxHvOkfsRh+vsp/QM7DByQg8/+do9gm6vjGjbC/Pqg7899ux6MNK4Ft44bpd6dtNAM0DWDD10IZUxCqCwNiv9dWJvdBbdnA4plsWocrQ+PNO1C893DSsnew/geq5BnuuCfx3wLp2FoHq2l62LyJ2wQocJe4TqkjRqDHbOsntvYKnTxPLcUnDueSrlN4uok9gZ5pQoWbQbKI/TKGeKn880TORJFM6zS8RTEW2X0Lp3/bAzPeJcI4vIW8IDH+KMoKyHvCwVv3DPKsVjZ7LGrtQ+feaF0jMERC0ek7nUKszanqYJsifCMe3VQ5ZbCRG01bzb5LjHmeIrESPihALpp1JJOUJMxTcRFTYbicgEu6mk1MkDXicwu3fcO2NrqqXI8pTqDsIYUS3nyU+d+qT7L6YbPZP7j2S1jYBaVFDIXYnO8YsxGOG2MbPR3LQsapAodHG+S9prt1OpE0xlfU7pzIq7TmeyHV9TZ4NK78Zy5rgn29sm6Uhm/Sn9J4jY7+muUSqullQzO+pKuWuPnH7Af3IKRqldC9Ot9tzVujQRQZe+j8Cqc3+Xe5ay/+lGeXYtIiFPP82Q4rjEf3/vmagsLo67iHgaq4rZldlInmNcVzp2JauKGLUPEcNBwSrVi/o6dS+j9HbF0105Kcybj9xnrTLa8vpdDlF5GFHoyl0D5Y7jF5/IH8W1EF5Esa/ghRuKz8lOcBF+JNufMExSLq4SEdwcj1HoPCa9EbWGU/LEOW2nuASUmlucFiMtDBknMBDbzjjsf7pDspOIbI3cScDqVqTXM53evGumdrWOV1YEUuwf41nh4f61G7lSo/hpklJREVmb85+eqLqr1b3uM0vlmR7/ycl2u14pLfJsy/iuBnelTtizOqeNqCfhuu1uVZrvOn7cJyTTlotsGcl0AJOabGez+6TGyauKTKXqgWlscTqMqIzw2nek0os4Uc3rPBDneRdeJxmXOrFYasjpnD3sdVoEOy2WxhSZxHOev1V8djHT9xmUOIy2JCazRD3jY5Swejtp2QwTkVsWa+CUZuz5ELeGvMen2pXJ9zRx/m6ItF1VnpO6k9pM464RS3k+k1UjOa3w/8TaFr+SIXPQMqLqzQHFsWT15CURi3b7F96QCzjjGA1max9X/JFSO2VkFO/QaeySnO5Zns4dad51dMWnNpuccybb9VU6jjp1RFFqAhY/TxJZL00xXHwGsZv3BTaHbOvdcHjwzp2NgiqZXBan8bViFKdY8rijwiZxniMh9Y1b55byhEpVDjLbUNTPytMVRc1XiohMg/OsdmcFvrrGdFXcvyriPtW4+JQsp43i9JL/L2wCoeeW/9d377mxLCi+1B4+r15K17WIymyRI5W/fMNOgXSuhuD9MX4tsqoyguK2sL339kyvPsdthcv8bIzIaEzCQb/OXdbTtWuYRHRGaphMUDlTPX958spZ9YQw0wNysRNPZpfcpwSxNjKLXOrMHYEBL96e7mlAk1l7xeeByd9jctfr/if1uaVyZt69st0boobVqUfZiUwp294Qvgmr8ok6zh6VNoPg38OmpalYH+JMAWdt2nQrURLcY3PGosXZqzned1yx3KeXsxRRndLnpksZkuQ2F6PoBrKKLME1jiLHCZ/UnSFjNVnilwd2ZTcAq+11VicaRfhkNZEJ9AVHS/h5Jp+v9OPZKQbTtaicOcpiXSy1hRo5k7+ZJn1VQYNCppY+r6Z4/4SUW8IPz5LicznpWsndtM7fnxN5/UV2ZNDLR5MxRLWlW4Y54d217D+FZZhzmKAT9cQ1AXNWkUK7YmWJU62lHeKs9E3yDKoTKmf/Xr8hKH4HYTq7bab6RVIpLaeSH0J3MxkeJidgjpPmq/r5r9K24W/YtmKmq5Os5dlGlo1kmbY+zAEO8DxXqKgfkSPU8f1e/A3EMP99QJpXo30P2plIercFvtNcmshbd+GbnXa2l2YgQ0TN2an9esMls4g0t9eKmWb6lGYCO/HdAZyvgLUd73TbcxBZSpor7MX+BzDDyn7/rY+485ZMerPfqmYjdiLedBynF+GmvwvMYBMzdULOZthsX7GMdS/2p+al2cymkxFdj3PtQhzT74OYcxYZzz743mtjmc2X3afXi/AkO/pblSH7aTF8k43Ofg15PWJ2DTxlq9yGv4Mc5jn1Rg5VG47MMKMj9vr4byj346p18+w2HWeNTZFTYZzNugufiAzwAI6wAeHR7VXuRmro4fPqU1Z9A85/Pd6XT1n+vJXPvg/zwgMKnnrwXarbimleRzrrQLg6MJNM57bAfltCsR5xyFag3QVrP85RzDpkQ0PbHw+jhTk3iKc9Ntz0nTb+nGGsDesMmhGa/hKcqmN+ux9nxSojGM+w9v2YbWctZaVEi51NF3xVTH2Cqpu57BA1G5TTBoqofZD/kur0eYDWRoRxhE58h/FeJ0If5nKKjd+G9QJsNcQI/Vj9sQbn3INz6HRIENVfcEpX7z1ZUjcJq15YocUVqqod7H2WUgJl/BBadcI/KD7hiWWq3DVpk5/lUxxtZ1nMiKfOmkyDFusz9Rct2N7RyWqfVMin0oQHK9mf6ZyN5YU/tb5Z/sagOOdVtf4ErtjutjxRz9ZwWgAUTlFjNV3Lwx17kLvwvU6+nH6/zjkU59OLKYVR7nTwOXn+T/g6K3hNsZSmVKeuh89FihQb5HKdSZNutE8a+YxYjHaySmTnL+wati9Raiec6m97xbLVehb3qe/S8qUrxWpF1UiTiD2LuKD4XUjDjoOKyPEor+sTtXqqdyI928y026q/eFPq921UD0D9Pbss9+FEhFLoncmtrw2kWIvpWD/WAd8GiFpnx95vwXcEBXQ61ltHW0FYC50oyYsryLq5hmGjS2+SarUNnhAym6oTq/sa7Z76+G909/Fee4relbZln02pbagJOjnVCr0hsMUq2NrwTjvXfP9dyEtXVJf6Tb4hZa/j1LuIhSyYLsU569HUKJKXH1Sc0ZRxbLEfQUQJlrmiBDLO7zxBZKpIpDNmMr1fTGQzGCZif5Dwv517P0vtH5U7RsWTbsLqfzYQmcsfhFlQi7iZn17n5l5R9yu1lPuXsVgdi6j0l1Equc9D1OSkXHlZ4Vt7V8ioZ02o1cPOPJeASp4cxfzjBJfOMlcosCTjAFni3tni9rJ3vRf1FyfVGGspvTEVDbEYmYwCOO0lGbdl8r3UKMUZM0kbAs8xXAe2S0NEIVl8W1aW7SptUQs9VMQJKaL+nqdaU8N27DOdJjhb/X1CQQPu2cg8r3O+k+GDrgqr68m78OCM6hfD7t7xz8ZiFCFPShS5LVavUtpm3zXJYSo8OfmuCuf+8W6lJYWulYg9msmiHlo9ONy9E6jT5vji/d/9RNTZsP46iKz7KV33Jm0jlmMQltsq167liH3uVgLbDbuwNGFjWZxgyTLJ6gkKMlciaxqnkzVxZj5kTtq9c6E4I136DKfJfmGAyeLpZTlULE1Ow85zSZxnaTjPC1H5wms3o5QjpbwCrzMVWS3sOBEVKLu699HpOeRtOWjaGVJ5JhrbPZLiFqfMzcrfbSiGUGT41DeclZOql+wl7UUWNU+SLi4VMW/hUwu6UeU6k/mn2xpOnO2pek+T0aqo3NVxR5/YOVNq75U7Pz45Fzjridg+UZHnE3vqZI8y9yOsMRFRjyBPCuvQqV0Mwn6RL+mQr6IWafKKVnniZPG5StJfEve9s3DFvRaP6vw1aifFS4tTWDa7gmNZR+ekYtXWmN5aiV8wMnmGSVhnsm5E6ni1xmcUOUra9KW1qCrB2F4NA/USy5LTnRtxIvch59GC9aKAkII/dXe1cyeFt7U1fXtaWK5xW+czuothjEueM+D8XW3Vg1D5XsDrnLf3ahVb9c6YyvT3jDpzbF5jSfmuQp51QSxX10m9U8+G6eWl/PSiqeAt9RsessJHhUw9v0b4D4KfZYW5epak0JLuyFUxLTC9oNZwi7OoS/saqp5Qf6/IXceo4y+9MN+49Dkzzl8ayuMMRQ5PrVp1n5FXLCHVinSZp1Ul82QnCrLKK2YVRciIS2eK0Z1SU7U1mAUsK9sZPare29R7R4rH27UTAUP/NdordZaEqs+n+3sGag2O2PcgTnlgUT7qP05FJ3JHhahNZzJDPZnBfR6v0Fjy9DJxHpYq8ZMlKSNl579VnTadeuelfN+y17lCqn4QfjuT3jIyFFPoTER7vPens3URPnjIftckonpTnnnF4rViR9vUNlfxeagxm4JSthfs9IGKa7QEzbp/EWpikup4Z9ZD9SS86F7iS40LSdrJ29whda6gSKPkW+6YTukIXQohYNXQbKUEFck8E/PHY67VE961rLFRec/rfLYM6mWziOeGlZzUdG3qXdnJ4fwlt2Jfzm0BMJyJfabO/TNixxrjFXGGA4VR1LXEXTOUEU71t05UH1ytABJRfGkvhoi7/lKe7jmGK0bpepEdmZARtF2tARUybGqbxUSLPm7rMpYHE3VI4nTDHLYdRwk5vb0pdEZLMIK9EutRR9BmT5PDyWL4bzp9LHatfgeuI6vsYfKyn4ja5FauCfF/5exf2d5Ew+/7wzcNv/u1HfC9ilhkFmtqf07+v1lkb1JD9iGfIvuS/chsUgt9ziF1pB6GnkvmkQPIgeQgMp8cTBrIAnIIWQhLvQgmshimTw+PWEYOBRQcRlYAMlYBClaTI8iR5ChyNCwpLUVo5WUQNCnSSY4lx4FoWYsph15yPCa9BsggWUfWkxNAmH+anEhOIieTU8ip5DRkyggP0A2jaDqdbATUjaKQSwMxEx5MycOSjIPYnAChcCY5i5xNziHnkvPI+eQCciG5iFxMPkMuIZ8ll5LLyBZyObmCXEmuIleTa8jnyLXkOnI9+Tz5ArmB3EhuIjeTL5JbyK3Q923kdhtPdziwdie5C/7eTe4h95Iv4Z378O+Xyf3kK+Sr5AHyIPkaeYg8TL5OHiHfIN8kj5LHyOPQ4lvk2+Q75AnyXfKkdql2OfkeeYo8TZ4hz5LnyPPk++QH5IfkR+QF8iJ5ibxMtsIb28gr5FXyGtlOXic/Jj8hPyVvkDfJz8jPyVvkbVJRHoI27YBtH/HDrO+C8R+CER/XVmjnaJ/TPi6rLVtR9oOyn5W9XXdx3d/1Kn1vfbZep8/T5+tL9GX64Xqb/tW58+YOzD1l3l4HPl+o+Phj6E+Hmd8Dc/gGwP1tbRX086+yfaGf75e9UfbzuvPq/gb9zNL31ffXdexnqb6K99M/90Tsh3z88cd/Z8j6+Hf8cx/693+/Tci/P2Z3fvnhu7exq3cvfvda+HvRu+Pv7vHOZnrn7Y/forOndN4GdLMOPk8gJ2nwqf1C+wD+/pH+0/6i/Yu21nZoO8uwpzJ7kbR/aTvx83/hz1dhneg630H2gLW+FVb5TlitmwCf18K6PwX4vAvWbibZjewOa/oArMRzMH9KXbcBff0QKOx5WJuX+NpoQG/bcH2+BnS3F2DrNVylTSQAPHgjUOEE0OHZQInnAJWcC3R4HlLixUCLlBLLgBYvA2rcApR4OdDFPUCNVyI9VpJqoJH/0VaTf2qHE0s7gvxLO5rs1JrJv7UWUtCOIR9rbeRD8jetXFujES2slWkdWoXWqWlau+bXjtN82rFapbZWC2hdWrXWo1Vp3druWr+2m3a8NkPr1WZqfVoQqOkv5B/aHtqgtqe2TttLW6/trW3QZmknaDXap0GqVGj7aCdq+2qnaJ/STtZOIn8mf9f214a02dppWq1mUNrW6jVTm6sNa7oW0+ZpI9qB2unaAVqcUqLWoKW0g7Wkdog2qi3RxrRFWlZbqJ2hLdbyWkjLaI1aTluqbdKWa2cCha3QzgYqO5cEyQyyQzuSfKQdpdVpUe0gbaO2QEsD5b8N1P5j8nPtUG2z1qSNa8u0Ce0w7Szkq6eRw54BXv4e+YL2Y+117UrtGu0K7SrtaqDc8ymfaRdpl6D0pHT4TZBbHv/TSEUZATTu15E3MtG4kew3ktnWVDK6qM8czieMDH04f8ce5KMm7aOlZR8tLd+xb0VhfiHxryv/dZ9v//2qd99/dt+eD+0f+8Fe2r7QE6zlDJCp80BGrgY52A0y7lRUBmfB6l8F8uZewOKjsNrfAdh/CFT1MqzKz8gvyR/I38gO8r+wcrMA5/MAl43aSu1oWN8uWKcIYPQM7SztMzCra7WbtNu1r2iPad/VntW2adu1N7V3tD9o/wB+DZTNLPtU2byyhrLGssPLjilbU9ZfdmrZSFmybHPZ+WUXl20pu6HstrJ7y75S9nDZY2XPlL1Y9mrZW2XvlX1Q9teyQrlWXlm+e/l+5Xr5/PJQ+Yry5vJjy3vL15WfVB4rHy3PlW8uv7j8mvKby28rv7/8ofLHyp8sf658a/mb5b8u/0P5X8v/XaFVBCr2qti34sCKRRXLKo6p6KjorTixwqgYrshWTFScX3F5xQ0Vd1TcU/GViscqnqx4ruLVijcq3ql4v+L3FX+pKPgqfNW+Gl+tb67vEN+hvhW+w31H+9p8nb5uX7/vBN+JvlN8hi/qG/YlfZt85/ku913ru9l3p+/Lvq/7vu17xvcj32u+n/t+5fvQ92+/zz/TP9t/oH+Rf7n/GH+3f53/ZH/UP+rP+8/2X+S/wv95/63+e/0P+L/h/67/ef/L/tf9b/t/4//Q/z/+fwcqArsFagJzAgcGFgUOCxwVaA+sDQwETg6YgdFALnB24DOBKwPXB74YuCNwb+ChwPcC2wJvB34b+GulVllVOaty/8oDKpsqj6lcW7mu8tTKkcpNlZdW3lh5f+Wjlc9Wvlr5buUfKz+q0qp2q5pVtW9VfdUXqx7NJ+NLljQvCeczqWzaiJin0hvLlh4WMTKpZKORyDnu0htGJJ8zG9OJ4YwxZjbmI9G4mTGz8SxcjhoR+lY+wtrkI5F4JpIfjSXMTfSLwR4OZUz2Ygr6ipjJHFxn4slh+MjFE1H6aCSfHDYy+dGEkc8p3aUTbTCAkQsnhzuPpSAtXbYiinfM5HD89GbsuZlB0ZwaTiXNjc20Z3y/OYwfrewvAtMq4Wu1oWkVE2zDNmHsNMyu7UZhNkiYDdKBbTpkbx2R1OiowZp22C+tGTIya2SjTvtBJ77fyfrsZH12Ii6Olc2PU/rsQhi7EKgu9T502Y3PuvFZt/KsW8yrB0frUXHcw4busZtkE0Z2BL/1sb/YXZ/SXT/e74+Y0XgiYfRLOPuLGkGHA/j+gPIIl6+peQBwMogADaoADTKABhkuBukiDiJC1mOP6+Vw6wX9rUeC3CCfbMAbn8YXPi0A+bSNdANHNdhABhvIsKnFMPGDkR4jXYWaJe1GRMdRbGNipya7thuZbBCTDTKMbYZlb8MKWobtl0YAMyOyURzfirOe4qynOKIkbr9yumy+UekzgTAmEKiEeh9eTOKzpJFOZXOZVHrETGKzpNIsKaaYQhBS6jqlGDwpu4kknAz7i91llO6yeD/LCScrQc4WNYIOc4CFHPaRKyKeljwClFcByjOA8gxBebqeecTSOPY4LocbF4QzjnQyIZ9M4I3N+MJmAchmG8uxU4dip8bg//h56gj9E6d/Tqd/NtI/Cbwdx1twIwHwLj20dXk0BTPONKYSUSpv6WcWMJSgsg+uJ8wk/dhsskcAP/3IjeO33EjGxO+xVB57iMXH8Hs2vgk/ABH4uhkfHsnRi2ScdZCGNaBCll3mRlL5rJGM4tdEPks/R+NJfpFP5OLpxAS9jsbH4lHswDwjbyToRcLMYrPhjAkSF6FI5keHzEw2PixBh1sUdPigoNMPBB0uEHT6SUGHTwo6/UDQ4YKBDhcIOnxif1EzSfuDD9of/cD+4AL7o5+0P/ik/dEP7A8uWH9wgf3BJ/aXzQ/R/uCD9kc/sD+4wP7oJ+0PPml/9AP7gwvWH1xgf/DJ+0uz/tKsv7ToL837S/P+0qy/tOgvLfpL8/7SlEyWLFvGPpYOU22boH+YxMArRaPiV8kT+F3qVPouo116ZavzhOsNKgPwExmXXiH9sgv8yJjDcUqmZpR+OyNvZnNxsE9T41SXgyhMGKP8y8gElR/0bjJKO4OLUX4xnI8nskDniYQZy6nfM4gDdiNhjqZySgP8LhqkjYyZ5A/xWjwYAkGz0RTv8W/qQ1N5ZIoHbLpwE5rL6ww1NuhjXIIlq74Mf1Yualqq9xr5hN5iZJJmVgfu0VtHMoAUMNf1/sjIOIiozSE9n4yaGZ1iJJ+FFdcT8QglRj2XshtFUnpnMtK4uBUWAdgybiT0gYm02ag3JxI6gpbVQSaZmTHAtvAIdOoS6NQn0IVPcBe19MPg3W0lfwA/ao12pvaU9k7ZfmB5H1F2bNmny9JlV4CV/deyneUHla8EO/qU8mz5feXfLP9u+U/Kf1VuVcytWFtxY8WvKj70Lfed5TvXd4XvC743fO/5PvL9r7/KP8vf5e8HW/Ua/3Vgqb7qfz+we2CvQF1gfuBQsEePAzt0AuzQawM3BO4LPBb4TuAHge1ohf458FGlvzJYObNyr8pPVc6pnFfZUmlUnlN5SeWVlddWPly5rXJ75Z8rrapA1afA+jy46piq46rWVZ1aZVadWXVB1Weqbqn6etW3q56ser7qh1UvV71a9eOqN6t+V/Vh1T+qfdV7VNdWz6teUL2k+sjqcPXa6v7qDdWnV09Un1N9YfUV1bdU31P95erHq5+ofqr6R9WvVb9Z/YvqX1X/sfov1TuD/uCs4Ozg3OCi4KpgONgTXBc8JRgNpoKbgxcHt1irf27taQ3Vfzdg5Q/9n8KhhZUrC7sV4nUnBwpn/+Zga0nd5a/WFGZE/2odVv9CwDrw0b9aPmvGIw2FFXXHBQoLT15YKK+bWWPN2v5HK1T/lYB1aPuvCvsU9mw+pBCqOz1QWLJ1ibVv3eVba5qNH1jl9U8HPnzwtZ+9/uhgobpuXeDgWPioumDN33/07J/fiT24+kv1kRt9p960+dZHZn/3K7c++62HzznvrrqtgRvyG284vvaA4weajviG+UK+/pGzfU9O3DZ+yuxPnzHaMTBy4w0b6zoDm77y2HnP1Aa3PfbdrVfWd/vjn9m8aWPtxs23PvCZ+h7/miu/ueGV2mCNdeDWn1jz63eMr96ZChQCw6ctWTz0mOWr25FesXNzoKBvO9I6COB5+Xs33/lQ/WuBTcMnbVpbW9jnxN9aq+jce7f/8K/PfuOcicfqXg3cuvHUG3tqCw2FGUcUrqofDARrnn/k5lvvrX85cO5YZHygdvmpT1ifqn8k8PcHtr7/9P1nn/2Vum2BL54R/8L62sJuS8KFjvqN8MpbP7zllkfqrZ8WNrRZ/sIpgQvHRi4eri0cGP2DdVT9dwLW4sf+bM199dHNm79W91rgluSpX2yrLbTMm184r94MFK7606HW6rqZDc+d9F7dHS/WDKbve/7Ze+957nP1b1/vKxwaOHtz4txYbRCcqh29vuse+8JLW2Zf4d/yhWOvPaXujzt7fL/2W/pHnb7V/qadZ/guPO3ik7bMvtS/5ZyXP3Nn3UE7Mr4jgWTeWlg9648LqoM7VwTOzg5NnFbbdfbdj9Xv2GvVzr0Dqx4e+E1dsL161huHVs96vwE+w3Purn7q2Wf7fS/4V1iVvg5/8MZ7vvnFx2q/f1tLtfH1lcevBy1Qd4k/+MgN99z7/OwXT3v46NjGiy4Yq1sf+Nytd159b+237j7nxOBN1vqaE+/KffHB2V+549avPXJbfmN806bUOKzJ8tN6Vx7z8AlvPnjvF268rb6gXXraWGdtOnnN1efUrw9ceMN9F3299sNXvmetqP9awAqc/KPFdesqvn1H9LD68cBF54ydm6g9dezup5+67f5v1M8shGu+73/Emr15TtA655nCwiv8L7zhe+Lp+7/14uyfnfr4qqGN559Hobr+1nuu/Wrtt+/MbDjxnOqPflfT5R/oG3zOd5z/nUKlb5s/uK76p1a1r8kffLL6pvpnAp8976zLzq1NnXX9N+s/Klv176rAYY+knqsLtl09sKUOEJt6I/18XWFwx3G+uf6ZNe88/fTbvzz1m4dPXHDZpRfUX/N3n6U/Uxjwn1tI+K666a6r7qr9wT35wXXpZN9o8nPXjdefEDjv6WrryAtrnnj4trvuqL/aH7yh+kGr+ojqoL4z57s4df7xW2Z/1r/l/BcvuaOuaUfOd5h/pjXr9jGr/zXrrb9tGQ9+6Z7RR2offeDBl66uH/TnLzn//DOBFx6+6/rrbwbALzl74pKx2tPOvPHbQIj/3DbXmnF4YUZ3xw8L7wWWP554tW7mlu3WRdut87drL75j3f3r8musl2qeL7x07JzCj473W/MKL9VYF1kLtxcWWuf3Vtd879Hb7rm7/ir/rLM+Z/pe+ShzeOEC/4158/MGsFFHY2F1fS4w64nC2re7P6y7c2vN6ujz1h719wWsWQ89897DD15w7i11hd8Ezhw7fcKoBZkSuvq+F34521oYeOnOe167om7Wjm2BWX9puyJ390uzrVDg3eQLhdDVwPqvXXz/+UfNLhwSWPeFE36+ue6zgdWbz90wSG8cnTiv7WKQTi/7dxz90cW+Tn/w4ed8Qw9/74zXaq96taYvevvXz6i3bt1R77sucVVsaHaDkGeFZdZEzZ+osPub5bus7qmHN8XvqS98cWet7+L7PvvAQ7ODz5Rb759XY4VBcNZbSStd2N2qLXTAf7sX9i+kChlr90Kt1VH/nYrCIYeAuDzBOuH3VrkVshb8qVBeOKFwwiEFX+GQ+uCSOQDbh997cr11S49/ZM6a6ppt/q9Zx/mW+YN3P+hLfeXRs56qvfzlGjP5+bvH66111p2+q8avzKVnHwuSs+2oukKldWjNyyBRt//sgS+dn7utrtBfuM13yS2X3X7X7OBBcwpn/mGBtbxupnWk3+p7z5plDdc/FLBSDR8WOuoKRwABt55X/faO463Gnce/teP44KMXvmqt3mot3Rrk9PmooM+/AX0+K+nzSkaf6xX6PPfWr134SG3wtQtfHRqqDtZ88NTL1n5UGeh9zzU19XUV6upGA4UDn1/367qnK47f9N3fRuYs3nRyH6zMG/7fWJ2++wNL4O9R/t8VNviCCwEp2/wnWEHfUn+w5oVn77rjrvofBMaHTztzoHbxyd+x9rqs3vr6E35rVc9rharjT8qdlKj79rhxZ2dtzynnnhmvD9b81L/IOuF31gbf0f4lhU5fAgQtMvrlyOjP3P+tFzijX3DuWN0Jgetuvec6xuifHqkOLpozU9v6TvnWfd75qHOFP2itPO+1j3YbD9a8++S33nxp9L4Tr6+/PBD8YWLOi2MfadufnGMdsXTOoxXx6rEd618FjB71u3ZrhbV49gMBa/krVuMbH9S1+w9fsaoQKnTOzgUKnW8XDnlrZd3Ml/Zpqg7+vGC8atVt27HHqxeOB62WLVZwwdiOd/4J/Lrj+lX+4I6FgTu//PCNj9Vee+Nll11XH6x/NPDr2x9/5uk7T11Ud2ogCOo/aF2wxSqz7rBmwzt/PfH7y5addGxD3fkw3af9t1913edvqQ0WVhfu8p1iLQ8EtYfet155vzxovbu9cPlH2gr/zKesnzy11x0ngvi2moKFsDU7eNicmRe3V8NLg9XBTSfW79w7OLbj+G3B+AbfQ7ENdx5bWzgSLYP60wLH3HDkV9fWHVMo9/3YH3zH/wtr0Gf1FQatv2wv3GJdaRnbC4Y1bt20fec8WMJV/8687b/ZuqXGWlq4xfd7f2GFdbe10LoYhOfMmpf8z1rVIIoLG7Z9tM948Ld/XOyfuS3wwu33ffv2my797PV13w9cc+6ma+O1R67fcEB9c6ExEMSn37r9ZvF0HJ4etY4+Xdvxo6DvUH/wc3Osm98pf2mfd6yfWjcXfkpXsvBk4WbrSVjQ2+e8bM1JVQc/Js+dRz4ma88jW8b3en9rEFTHU39/Vrv7FWufreXWyqB2/y+sgV+WB7W737TmvFUOJPSq1ep7OVB41WrxHRawzi3MKJxb2MM381t3XpzJZs5de3F90Fpw/nYr8Zr14HZovsdPnrT0+uCFr+yofSVY816gcK410zrX2s33LvRRaPF1BmYOxRLhS+pn3n7GnOCFz1gTzww/Y216dq8vvWbVvGbN2G7t/Vqw5ifWImve64V51sJjCgsLc1ssUB+P3p04oi4dCG6/fU6wZuuOw0b8wVvmzNzxuY45wduB+z43B6B+6Z3gTdWPW9V73f3HQ3bUN3w46wOgzg/Oq34LRDJg3frbK4W/+QtnVlh/exUuZu5YGbjxnkdBWz8fvPXWq66+vu7JwKUXBS/dsm3HzK3BvzS8fPrLewXf9b9k3eILvl1/b+Bnd3/50S1zDs/DNE99pXCqH6nrS78ALL39PpDuhVt31IMU2e63Zu3w+b5VmOizllxKuXD9S4XARyNW4KXCev/Mv93+RHXws5+9sS740ZlHVRcaDj2qOmhFCoM11gHbXv39B+seP7g+CPcXw/2Z2n3vBK0Hzq+xjnwLbdUnA0H/2urgRZ+eE9RufvHPL5Rb4eDbF241x3ac8Epw+0fa+DVzguXPPjQnuM/bxpzgxd+stmrvA7Ts+O13qsufWLyu+hvWgK/RH7x6zu+sOdqzb5cHH/nNI5q1+2NvP1ZuRYPandZE+VPBx+O3b7iyLmh1nf9qdGzHr/7a82rwguesK547+3nr8uf2Cm47+5nMN+qe/ik1dZ7oqLmqUCgs9c+8b4518rvlwdc/AAvuQWA7a8075WBKWTUrArP+WKgJbr/w5eN/bn300+NfDn71o7xVqb1vnVceNAKvXe+bqZGbCM1rXk7I5wh5ipC3CXmekPcIuYXmPskPCHmfkJsJzZLeRWiW9F7MgGp+MkJI1QpyFiEtc8gDhHyNkIc18m1CvkvIs4T8kJAXCfmRRl4m5CVCtmrkNUJe0ch2Qn5CyOsa+alGfk7Imxp5i5BfEPJLQt7RSBrTr4ZGYoTkCMkScj0h9xByPyE3EnI2IRcSchkhVxCaJt9CyFUauZaQMwk5jRCTkNMJ2aRplxDyICEPEfI9Qq4k5DpCUoRcSkickHFCmgmppAlf8itCvk9IhGifJWSjRs7QSF4jm2n+kDxCyFcI+YaPfEcjTxLytEaeI+QaQs6pIBdVaJc+qB1DtDaiNWtamGitpGxlVNtBncNyEiBfACz+nhS0GVqdtlgb1V7UPiirKDsYnMPtZR+Wl5XvWb53+fLyzvJbKk6qeN2X8E34LvFd73vR9zPfb3wFf8Bf50/5r/BvDxwRyAReqGyq7K/cVnVYVVfVKVWjVWdXXV71ZvV11XcFy4MzgwcFQ8Fjg/3BE4InB0eDm4LnBX8T/NeMihlHzLh2xs0zbp/xwIxvzHh/xj93a92te7eTdkvsdtFu1+/23d1e3e3t3f42s3XmupkjM0dnXrv7kt1bdm/bfcPu8d3Hdt+6+692/2CPrj3u2uOpPV7c4/U9Pthz9p6hPfN7nrvn5Xveteeje35vz5177bfX0Xtdvtcde72+159mBWd9atYBs5bM6p21ZdaTsz7au3rvffZu2Hvt3p/ee3jvLXt/Y++/7L2zZkZNXU1zTbrmszW31jxW88uaHfss2efrn9I+9dVPvfypHftm9n1vv6X7nbXfZ/e7Yb/79nt69t6zx2c/N/s3tdW1eu3RtXfWPlb7Qu2fav+yP9l/0f4D+98z58g5x8+Jzdk059I5n59zx5xH5nx/zvtz/lm3vm64bqzutrqH675X93r9WP0l9V+s/2r9D+vfqv+tXqkv0lfqx+mn6GfoF+lf0O/XH9Wf13+t/31uxdxZcxfMXTv303O3zL1p7pfmPj73B3Nfn/v+3L/PK5s3a97seXPnLZ539LzBeal558y7et518949YN8D5h9w/QG/ObDqwNkHLjzwCwf+/iD/QfpBSw9qPyhy0DkH3XbQgwc9fdAbB/1p/rL5LfMH5p8+/4L5z8x/ef4b89+f/4+Dlx98wsFnHHzRwTcd/K2Df3TwBw2VDfMbuhoSDWc2XN5wT8PjDW82/Knh4wV7Lth3wecX3LfgOwueXfDGgp8teOuQwCE1h8w/5KhDbjvk+UM+WFixUF94+MJ1C0cWbl54+cJbF35z4TMLfxmKhm4NvRR6J/S3Rf5FcxctXXTcohMWxRdNLLq6saqxvnFFY3fjzsULFjcvHl984+I3lsxf0rDkuCXGkk1Lrlpy95LvLPlZE2mqbzqq6dSmRNPmpuub7m7a1vSrJmvp/ksXLb1/6RNLty39xbK6ZbFluWUXLLtx2UPLXjl05aHJQ7906DOHvrO8bHnt8tDyI5d3LT95+VXL/3jYPoddcNiTh/11RceKS1Y8vOLnK7WV7SvTKz+38vmVO1cNrbpx1WuHlx3+mdVLV1+3+t0jmo4474ibj/jFkV1Hnn/k60cdcdTJR/306KOOXn/0l49+/OiXjn7r6A+P/t9jgsfse8zYMdcc8+AxTx3z52N2Nu/VfF7zluYHWma0HNRyREtry1DL5S3XttzUck/LV1u+0fJ0yzstv2ttb+1rPbE12ppq3dR6fuuVrfe1fr319dZftv6+9e+t/9sWaNujbW5bqO3otp6209s+23Zv2/Ntb7f9I1wZnhM+OXxv+E/tR7T3tqfaL2z/QvtX259of7H9447DOlo7ejtO6RjtuKDj0o47O57o2LrmnDXn3z/TWn70nJ+Avt1/25Zxq+Fd0Ov3/dLa9N6srX/d8WHN3Sda5CnfrAfv2XhH/Baz8jP+WVuv25y7biO4482NheX1+UCh+SerrT2s2dv+bDXXgU+ytNfyHfvTuh/dn33iSDN/+9gXN92w31z/rJ88mHog/eXRSmt+YWfNGYHC3OHuo47d8HVrXh0YfTMefe7Nh798/rl31H3euOakdbNnfm3Lg3OutDbW/Oh+X+EytL2s0YC18m3fjwOFlat81i9W7KwMrN3om/n/kfbecVEk3cIw49Dd0IOjO+2wK7M946przhkjZhQVsyAqCEZAxAQIiCBizglFlCCSJImIqAgIBiRHRQFBxLDK6q7renqegvWr6kHX3Xvf+9339/4xM91Vp3LVCXXCoDGhbtqN7pKCWukxyFDevUyhxlqhI1NwjXrETLejINQUPWLmrKTkp9l4MNjOQ80QfnaL70+sfAUhb7483BrJzm9+1ZmXNy5gC4CVwKJ6aV/2hOAyl41xEyndceGn7mytNuIE/xyLh6PK4HQFyN0xu+T4XGSXRoE+DczDIugExtOLkLGmE8iH03XNNko4/SvzNG5Nr0FrVwzXyGvRWDewKBF+KIe17pLsWnCql2ajecpaWE37hCR431JB+7IamKCJY2DKuLxu8+3d3Derd8MxUxqmtAxVLlkbFOitgWdM4NmEsxmqiisrkInGnkGc7RIztTyIvQYGfzMu8GkIz31Ee1u2KEex8jN8JagUUW97w2Mx8/FIFmeuaq5Tjmfh3v5sKMjexgujy7maDIcJfHPH5qkHea5pJwsvWqTKwTyE74deVhkw+M/PevnfMoBgjdkI7mAjxGL4kztZl5beE/jIlqVYduRmvstkuHhRSmM4yzU8l2jOQqmuIqMczz8zM91DNrPQhdQQGwptIVwUFzRcZKgoMqgPMFxUp/68hrvUKZ7FICNYb5Q3gY/SVe8YSuoPCm1twLEetxBUb84iE2SkbZQ8r5Q+R0bKSuHH8XSVfqXQbjwtz4aiCkkK/EeaAkVK+E8F+g8tfymcd5c8gN+lWNI0XcoiNznqvr9JeO2uqHg1pIYrqYMlC9lAwQ8zPlz9gZZpSnd+4h72ZlVulTRLOHaBhR7I4gIrL8A7w7UMjMthibskROgpTRNMMSvniYypUZPKbrBNN9iasol0MBhTT8qn0JGwBMvVG9ASqhd98evzCBo0E5VNYqletPwWagOjphUudQtz03YE5XT3cHdF4uPPej/76uV81mvnq3cn6QnnHwrdURtlDs11CT3xnoJ2x3OrgTYB2bzrQ2PVAQcprmN9OOy9FhDnHe7bkVPWFzLup1zOLDtqgDOOFlBzaW5ifcBMCrXdNWMikpggyf2lla7qswGbUMTiQ7ZnVoR1nMZwHUMjt6Z65ewygKGwQDnJyc7cVx0OB6mcTSHOU0yQLbxSlsZfLjyidkfu1MwLW2Pum8ih7jofWQF8xRvyxWUC66Aso4Wu2gBqIo1+azZUvr5Z/qcmiLFlEX3L8pX6wF3l0k0xrZzyNcIpr1umFto291a+ShcBuTpblstsBb6DgWMJMFenA+cyR25xXoYF90F+JSvdtO2LFFHPvOtA9TTyIZcCAX7KvMzw5BQN55a0iHpLm4Jhwv1zVQfrOprRey39F3ktRx1hwQ9cXM3NjOsvTJ7Puz1wkPn8LrMuLotbquZSRtgsmdEH76yKcdBJva9EaWp761l+1LWsBA0MR56U07oVnotVXNxcwVeZcyE6LSViw8Lpa9dbrVH7B1KbErbGpJnI0Wqyb3S7Rt6FYK3UZ1KwR0ZLefEN+tRLjyCjJexM7TR3yWc9f7zAUpirnaa0pA/1HZfusqjWztQu2qoqfXcf6j69503V4ugM02u1V10yxy6mdJsbpj+UwnQYuo9H4/vDfEk6tLnCy5E+7CyTZMJoaSbsVMLoMjSalqPR+5skD15JHxi788NI+wchRnoQHxksgCzXvsFrJIfevmVCaank6gu4+FwKYb5K6PYY9GGD5jYD3kPeIDXqNLQL2qq2YlDAi0HAqw+UKbuueQ3DNJUMDLzc0PTrleHIVG3GoOFrh3ZTy3tjbG2JR2r8VArGxk+1lutoNHQWZcOA3ilKnrK/XMiukET+CiMapZHGgMrRmUM0Rgzdqaq32VXA4F09qxIppttsc1+tNmeCYlNOZ6ieJ9oOH2AzZdZIDeqLbCj5Zb8KOFsO28oVuKea8IYpz7mPMA4LQcrqh2AMetbpo4M1hxhOKA69mvaLCeiZ5yFqzhJfT0dcZWB0/KlbqoeproNH262YqOE+LpmYRddfXTGo1/yVg5YsO3POUTOFcYu75l2mkr/AR358mbDAXXKqDixrpZBgXAvjfy3rKSzA+zu+eRk1Ey4wyLJMeYgBC0inDqNMWp6Ehd2r5bC4HK/wCF+9ys964b56jVKxh3zlC+gI3SbnI+NNHnv3eGtOgy0FW8txOTf0iToaefnoFVVTmkPPbsuse1jZBwY7aqYyHrG3thWq5B99yyC9ApaVS1IbwQuv1vH7yqGrnkJXTTwTFJ4SkqWqSlras/eqeZM0k/RBUfoWVKCcWY2YWTZuGxw14HrkefQrVUqK385QTSFzZovdicUq1AUZjUQxmrkMMi9BMuiilh8JJTc4imAQhtdx9eXGDTRXAo5afaoECRMhYjcjhzZ+ZUJJqSKxzrIe1tRNrecaYT8enOrR049NC9PHXdFwVffjMnNqTX43L0TtULup43oPvb4wd42aa5ywes6cvhgFPZwMP5K9NHpxWmnZ1fR70Zq3y3LWjjWZsGjNBDVXdRGeKYddsb2cb5J7ObG4PH655SyXlQtWqvEkVMDSCkgvk1x+Dl54Vi3wlh34Zw3EaO4wMGcatEGdkfHkAagT6pA/Eph7N0MvxqmR054JjkhP5bDs1HE3zTTGO/S6b7rqLrxXjltx/df3SZkPa5PGod6atYzXZgePhSo5GrK/CeyaJAWvMP0ndy9gp+/OoyH4xMHSJknhqytYnMUpYaAvDJYk460hgP4uvgjmlr4qkWTAB2kGfDzVmg2Xa6W3QN+Pd8Zk51KZ4vWroVX5z7mqd7BEWXWLTg8+ez1NXZgfHl9tAvTEjM4LbXdsc8Jb9WRs1Il0VVHq5vFztm1wWKGxnO26aZIJ9zoAncI0SuiKeuWCUa5lIZTkgzpXkVb821OhbTFXJ6hQL2VpZUh4Gj7F2xymbjFVoel9foclsAMoMAU7vKmRPhqFei/rO9HJL+zKUU0mlFET6IAN1G763FbnMytViBr3E5qp4TJ7p8yuvpkYFRehPkxzdatORG/MVQF1sPY3cDCBBT2gC1qDdqAeaD4KQ4eBQwysV3OZjnBAae8eVfJHBDDPLkX6+ARqVqJDlMfWdW5rVPLRvOKU4MNV9Wa51z1Z+Z/7y+BoCXiUKwrqYOkzC3ycC4xhKg29C6pABlKrtNGnNIfxea4Kjc6pwufZKh/1UCMzuq6ljRKOPme4j43XnQcMWWPbXTMYHirB4zEGrk3ZOHSI9bqRGvlBVpKEd8oNv2k8DAthMVsX8QIv63QeVyBRCtdnszPYo/wyXp6Cz3teOfiXH+UJ15cHNuXIBvzhaDl6Qcv9+fM8ZPt15zNZL16+nrfi5Y47K7QMxnbL66TQYec6FgwW4JoEE+OtbDxmcAuMHXm4vv/NeDfh/QtFQZVgUs15FBhXCcPowt1guF99iN5/AemfWq4Gw+Z51DMaZII1dSY79M1+E5y1G8mmmzQPr9ZOQx0YRDcvpbg0rxk+Y3QaqSqvfDUyEJZSSEnL7fgDLLeEVG1Dc87DWC54OG/FruUhyU/ZiX+GxxVdBpHleOgrGkWONvoXGhRVD+FH6DwlH/EaNJGuQ++UEFnLvExY83Pf1XYDNHKYpZWXrOWFy8hAGf4Yzcmgfk+4/RuMN0lnYMvgT2jwF5WoLYNs84Z8WqV+fI+oRD/AcE0+Az+n/Q5SNcj2pfRAw02mM6jX0h5EKXqBP8cWGPdnT7OjWXluaOuFsbnxWtaa3872Y+VguICVnBI2YdqJ0Q9Ya6dRI2n0fbMXtW3hDgdx+DuubstQo+8FL2owLXdiFWe1lpwr9DfuyXNJo3gwwhWECCukIcb19G9gTYVfDU0V5/ScY4S1+ldMRV7QaAo+tGPgBTWMHoqmUm627vaiusjtyuY09QCYSuFTPxm9oojY0CozYBZcEtEENu+kEcaY5+6FjAq0awokBdXSAmQUzg8mRDYYoqWwEVPZMhrZ6ajsFzjoXC2NQkbB7H5hKfiOqlbEQwkXGi8sXchydo0wgOJCn6MBVCmN2oMvNZWWd0JtciR7IV66FzOAEJ+D4ml5KtsAw6RgLtRZ8JasIqIeRtZyaaY859GZ5dJ68PJavxKBLyECyG0sgOCd6Yxxc7dbeIbHZA2Ya79163oscRw0bbGnfeJKfe6pDpcoV7ocPeqpAYE5cT7hbKqq+vJKLOMsZ1DXOSsmqOUBAXw0LwUf/248mvh4AnBqeTyP61fE1oNTDZcOa/zN+Jn8OFaODPfwMKRYUlUrrTKuFea8Ke5TROeFxLBrVMMXLe4s3kQzcgvWheU+dIrkud+yetKOPmcjzh4PPnlSszPkwp4I1ZvKbMIJXtke5RWqLs1H3WjuRacjHlsObVT1mjSri8abmRlld8lNzf0W6rjB3WWVycrwLdG+6t0MrtJmq8NqS5OxJdbQQb23QGm2Kio99WLMpUDNH3RuiKvt3DVre2u4X8wOCTLlolUxBZpQJi/uUk5WostUtRszc63zfLV8C0tm7ggrvW08jYXTprTci0Xz3Xhgn0uya6SnkfUYNkWbPo+NPcJDAJljKfyo3biZ14mpOTXSHOOTPOSHLuH93RV4BcxqwKy2w49czW3jo+zZqPigK6rMc+5LNc3KGmEEExyTeCZBlX12iy1OqBU0zIuliePn22xet1y9m050dwpeqLKy2bRxrWY3zaVHsVtYXOkRlqu6jad3BN2P59Ki2AP8RLZV9CzCK69r1qmOa7rdKnpuC03adkMFxvkvYDRmFmDclMLO85dt3rQRt8HVJI9hYGLzYqXDqtOBXhquCRqZoPNJwddV1XHWqK1mBcPVIMXKucPV8mqxjWJBXSQRl14KdVFKtL5GO5oJXjs/aIIKfT94PLLUODJo2rMFL29fDTkbrjYVxiovR+8MOK9BvzPeHo6etqox9inQQZPCQKfMa1VqeSCLO8ydxkOiuVx7njtdyXC5DiwXfXsMI18RwGMRvr8/Ft/7BiSwMC9gLG/JKzJrYSbe9vtYzmMW3vZz+YT9v4+0ZzFd6YTxwGjjOhr6CyepUTSX1B9VUz7W/mv3m+ym9+9M8ElX94dqT16+klecFl5zSWBifJbnSn5m5WFHeMlpob30tPEzuhA4KvzKmaj9Jgfo/YFu4UvVBagDwRo/wRn4Ho5gUW4OMnI2pzAy8ju/6YZ6BrQlGEONjlHyGLfWOwUslkui3oLxO2mU8XjWHMkLoa4Y6gsVp2sDq7GUnoLkZiz6ieychlJpA8Yapdo3k2k5hgxlkXw8vxz2lawpUcTchVF3OdcY2KO8S4cePhUUouKSImPWpqmuXY4rP6qZQ2/ZvX2bp4pzdV4fb6uyW7F+7E4NprOlDQrQ1BTVcE8xrX3Xk02Cvq0qBZyZfeXlFQWMu1Z7javD2e97skdBrRzOnotyY7e6rvfaulEjf4Gcb1VcSy63ddPKyxVJje9vnSvg3mFa+jPPPUXWJd3BaKN6IMO9TyPqnk1E3WPNgP7R5Pi7Jty7tLWhiw+przD7jh7df1Qln0E0L0d5LHy4DeXRopYm5UAWk66K4pXFmDqGnWV3VHGLzGCE1ns1zy32HcPjtxUst8qsR4u3kpA2TNhOmRHKdqUBk/RUPLBrmLCXXuSBaanBhM6aDPhB3RmcEVzNfejHcn/FaXd9GbElWwh8N3atEHWCb332EqIsxGdVNxZJhKiTrCXb2ICR6paayhqbWu46xqtbMV693oPnUjOFjCgMXBPER9SvuA29a0fd5iJ9McQqXwwS6YthLvsiIyFVeZvmLvqeYblYs0ksfrrF4ozduGgOmfGI+qN4yv1quXRS1puUTSdlb+qdFM7h6Q+OdmM9Nrh4eeLpDxCnP6XEgShwFNF12bfOk+knqBhPf+cMs+eb1f3w7O/grZnzeIWv8VeZfUeOkMmeSRR66TxMNmUDWlw7sbqEcewW6GjBk2dVOt8k1A1m77a4iLkvGnAD/njcrvXcf+C0AGNZ7pOoZuK0oG5ZiIFS+Num7OSWNcra5Zhb4nMwEjppSp+ptOCJxuhHojHaJtQpC+i00huTqXzaDH6kJtMBLavFAhhhsg2KnY+9a70fY4mtnQDKxw1jaGQ8RPzlXue1zBIBCZYDSRB/o3bdWdahhrtnduMLuos0SxnDcKVmmY1kku+a4UnO88WTfNcMT3Kp2W4+jAULU9a1xfUnXl4uKtzwqCeyLkKEBX+AjHo9+4GMOqnFBUMc0I16Px51AEaYEEpGrRvzB7ynFoogeMGj6zzPstvquALfoSzX4Is6CldI+3lkkSv1JpFO3CIZeJHJhaI/CzNN2XEtrlNZzhvjp/S5X9L3tSb05kOFVxZiqsqf/Uh6tL3F5VtwMleZtZtxz9bXcqlw/u+eXS9tWYght2LI63P5m8K4UrDCTOiUR1KYMkX5SNh5kaXlS4Q56TD4luTKPdDck8IYYY7yUWZOReWi7LHjFs2ZND5r/iNNur69y6WbNyNjr16NcrZZ7Oq8XCPfuaNwFd5sYDLJ/aI7nhqQf9Yb46vHHYL+SF95jz79gTpC378SfD5Sc5/xE026/GnuVLpgErkrfEfIjo5FjNcJ9xMbjhhwkSfKqJk0N8t/OoWYvUMmoykmTgzqnGnWsEU9HwKUVxgu8reYguoj6qXN7Shu1vLwLVmvTOS5+yvAPx+OVkpA8kx6DQqUBaPpfp7L569WzxTYChR5kElJDE5+YVIOA/LRAJAzv9+ILkuK2el7UV3EBLqvPL1A1WvwQtRDgxTTGXkCMi7VTitVhD1xeAI/P13yhKuCechYCW2K7oBelnfyqosaruJOeGRqnslL27hRseoxMdTyYK/Q6yY5sRcLMy77+ESoi5kzm1cRuzZqrkV/y5CVMRs13OvFmNuYYDL8ukP5evVDFyrbLcjTymT2yvXTLZ3OnFulnspwVZvjrvnmqrB8bCx0VdpucLHfrnY/ZnGBsjnnFnHd5PqF6LRT6oeuVJZH+CYrE/kYsAdPSTJ4EJMYe2XBrVsFhVbpFhaLF0+bnm6djxlyA22HLAnszivMk8JuTC3ytAcKWw7kaW+j1TTqNH9sD2T0ePYvmqu0/PWOkhVu2mHFmLSWP06uXl+N+bQ7O5TcjZcPbr0ocEqyTtDYxlDOIdvC4kw472KQZQBt0mh5o1vP/laIVzdn1ArXmIYrVwoyL21coh6BDBd2M5l5cU3WKjV3I9mVStp4xs3RxMFpo7WD6/FT69TTGC7dOyp6Z7pK7gml0dAj/lZ8fpTkfHzZ1ZTU/HgpzIBS5dviKmBBZl7Rq9fU8YhFbYvHN2mi9Uda53347VZeTfWtGd26Ws8YpbEsV159dJ6+NI66Rq98Sh1kzh47FRRjkux2brl6Bn11uPLWWeb80QAPzwD/rXhSkn1vgn4ehOWD3k1JfPanh1AwiT8J+sqr0YGBIXib7vDcvH2FytI1qUETw5SHpd5NuuDvG6bOY054rTu1SjVk6mL0o2YTg76/N++lWg5jhZ7VMDhd0pADVDqMzZHCCaGnMh2cDtJ3kRN1i74LTtRB5JQDKaPouWgFtYeeDyswr4HGoBRlDpYf05HN4jLCqsNowquPNsWoU4AGyS/4NEvhoABRfDSrAOEp93Gf8Saea7rJp/EK0RKFC47hubjrWJrbyHPBE3l5ExbeeuU9yofX+IwvrZfCUuN66AWqPKR6BKp8JKNhKLzGQswYNIFaO8ZziSjEeN5Y80Q9BsZT/XA2ek0h1gJU45BqBqho+TbYnAu/F0qCa+Dnx1Jc2WZl6qVzYSGaw/RxJyofXtINNqn9Fyz33eaknsscu3DpRIqqLMzNQmPPDPChhvosXm5qMihjWa36XK5y6aaIjPQLFzKPakqOU6gTs93b1ddRJY+Gzfnwe54kogaWYA55OW4CDAufw494BaDLzCfICNHmQ/BGc2FQp/sjwEh9MF85fXXWr5qbTENMXkneZZuu6sXMYKdZ5mp5LBZznsLoWjx8br3g7T+f59wdWW790oSczQ9VXOIcNosnqMqzjgsGD38lKCp0Rrdc3A2eW+KOxWFzXr4OGZcJl5oUEc9zXo54zP0Gy5DxZBZ1F1ybFOWvhtZwoeX4RL2mtwt+1BSay9r+QenOr0fTSoSdFeBUIrkAYdILQgdlOT23uQ812c9xq4PKw+PwwR0aS8b/TNjORFV5ZgrondXcFYZQQFd0pc8CQ1WeunQxQ0VsdoI1ecwRn81HHVXD5yxDlJdmK1JQw2n5yRyBlyQAxt6KnAU87EfTFvDyQJKYCI+liTlz+DxkNYeXg794nQv6r0D6WgrndfdStr2Ur5v60c29e+HeIjfM1uWDjFyv9OI3Qffj7GgdewXfnWW5R5jLe9WTRd9BP8xgVWP+6tG6zTF2Ku6V/WrnKbs1cis2lQAnNV48y0Y3crXQR2TF3mLIWsyIPYAfWnZhNkvouL8cDpWDezmxbzmLRcwHxsL4f5s4mdFo/hikj7Zo1jJo25tBMFl9sFKJ5DPuwlzNNQYm3azCYlpDM0ZWIPgXmZfC6AIYXEQWcmsdWeoMcPdXVmZevhejmRVze/ULVd7tC2ERGs7ni9XbgKU3oO1eDSRn0DBmbhlqN32Rywxn9V0Xq/hxKqSY0Av1wbIIl4EGFBJj64NFSjN7nbH1bwklVQ+vzkeGxNh69eSxasgSNMqS6KWI0qClNDAHr1zLVck3wpFCSQxkSmPgiBIyC8n1LqwSE4VYMVGILWyOxfz6HjYW9khjhWPh7EdkEc6SlFScknqYh/bIYiovR8sxzAcdiBw57GGv/SG9dpiXZ4Pp26yKW00SGP+KqK0CiFISE3/Je4wsurMwUpsexTcuYAkrz9X41O+q4V73ZbmP1cKmL4wtmuagDZCgboKfFHVzmMDLf/crh8n5T/NhWrnk8QtpiGCsvDtijkUuXpHzTXPBSHUnLSgsXlPKeDqu3DpXNXhhGvzgowHLfDT3KczNR7Np6Bl499ITVVbCdq8rGC7YZeXpeSpk+JM1snXXzKLlqJtu147S7drW90QYLW5Y+W3gSsFTx5dMeySNA04JPZDVI7CCnqWoJ3heZJEn6oHrGQvRbyUPXkgfQLTyxduBOGX5OLCQpP4hTR2HZ82cvFyFvVKwwK9ooDiV2l7X+ahK+LHiDfm6Liqf0rNgHLxWvvpW4fRavai5PT6vQhdRIwWoZdVXRZNZli1mtLJ0qiYuRw/VGv9LWYVzm9sKvZVLN8cE8+uWqbWL0IevGimzrFaVVHqWqJPS1bB0c2wwTzRU5BQa6E4hoqH7In4G0d1i0aeddktfHnVu+YWIPqP5AmB1UE7QYy47g73VoCio20IEtv8s42HVLZEFF9eYVCDDFbhot8zjR+pEp/WkvBV/NG0uu540SAw092i3KHPpm3cz5lD36XFgQJnTO1t+2Y0B8AY69HhbnRfhvVVaLx3v/cMX3ju/pWy3WIvMij997ThrRSyp1/JwgMhqHi2f8LG3Igp2nGRKxpDQ8kdrkgwnBZJeNYlQ1qRT21nw1xWErrzcmnQOJw3QFfzYmiTDSQdJwV9FKDQfi9/ZtdLsJF4eJ9SBSlIH+6V1Qt1JNk54F8Q/xBjs9EPhHWG8R2C+2wqz3SMw1316Nx8naEskn8BW+knQHsJvDfmSX2CrFFYIDcoupDZ4LoGRT6UwUoB5uLbKq7xQ+S9RK46IWq8VMJuIWPUwFjkrud+wTFV/jU+3xjLWbzt4TNBfwhkpeAh1y1hcYMkdVzcFsKWcHTijJUrRQlDDhaawyfFxcfGa2TRnt87Dxn6uyYJElwfq1kHV40FB/UhWHunMS6D3Oyn0HsTLI4Ua6CUBc0zf/YSalThXeA8aCQz8QwpnDrGElVgK7TDxF+pussRATib5Ezphhl+75Th+92qQvMa7RgoBF/FwtT7FONdO+qfWZz/up9aNGD2Mh5+lMF7rNpe9zTdqtyzib/PweAiB3vxltjYQ4G2NEhiFSfYo7bYQkut8ldc6/+tWgLSA6QRZEdeHIolIwiTCdQUpvwU39gsclYKn9de2F0JbKSzUum3BVGzYOB4moAnl0E5n1XuA5WpaL66bqndl6i7FpyaMUYNT80bKLmDDFhuV47qjx9w105jtEbE7L6s+pD14r9EuGNs8nxnOQ8+dFcsWs4qiOvCp49KgcCdmNJJrP2iEwOHNp5iJW+3NHZ0PH9+knsL4RV4KuKriPNC+Y7vFO72vF3ryeL7kzlzo2nqjmQAOeONA+77/NtEXDfRFU36iDXEhpvxCQxbNpd5ZyHAJDSMZLnUIas9ENdf/zyb8j5F1MfTS3RA+hfk1WPK8gqyVL+JyazTCQqbBLtn0i18IV5MyqvnkU4ZrCuRDifHxVGa+22ZbL+9DB30wc+Z3NiQgViUX67NnySVlTu3HmkR+36/Kp89H0NDtrw6t95Wj+bx8Ha45At3IqSCcgQILGna9+J4wAG8U4JD1HRj61T9CAdpSrklYkswGbN/nq3J1Ox59OfFyUqwGLdZOxowlWtwyjXpCZ8E8Clzo22getWGjk72NyaLLzrlq7iOy1UYp9wcuOLFWXYh2UmgyXQj+1MmEU7f3m7TePkVi9AdjyAVUrS+5gfLtybpC41B+EypRpicm3tLg9FurExYvXr3aWo2zrRLXpquv6G/0OhdFssiGJMlkU6rl0aJdL2GqPwm2BPk0B2I0ibnrdzf5gwSbtWqBvjoALNH6zmUPEqRGcoQuQ3i3lu3T+K8pMIzgq5stf5C0q42E12Fq/1HDQu1ufEb+T5WLuz2EIH6dykkwHMKjzS1blP/QPKXmE90TTEola/KFTLRgyDUYciBB/JmNx3hyzWxLTiRStzzUJWcVY1rhB3acsIznPlZm4RUsIAvYqRwOl2He7BjfUCoVBhkTvuwwLC9r5cvQu5ZpsJtx3LRi1UKyVA9O89v3+alct+DVTUqOvaiR+/Nx+ed5SNnAnmnx6c7783g+8PsYMhvpLX90x2fOmY9rtejGGf6D+Iw4Aqebo+9qizGmPUw6m9FSj6FnYgnmUTno5c6CuYqSOgiq45qChP3K2837iSV6EHpUJ+zPzp1PZ/IoiMnkFzCgRmOUoIez9Bh5Jouphxf/67VF/Hq+vNGKP5AZQvRveflW/FVx1tbzj0qt+OU5ShhVikZhFmIUngdtpbsisNS7jqsPRLTSKsHpwflzh4+cEi24yXbe4HECb6KS5LikxBi1dsrwFkuGq9+w2WWVjQmuUGSat9xOqIu/zdW0Oj0MysCI7h1GdDUY0elUf4Toito/jkzOiJbf1rFg6MznLxavOQfxMD5VSXSBxPS6gNx9PiIbcxrel448ThWsYTQGjBXClFEwmhpL70CjKdS/OQxnJzjzqQk8WAzikXGqDhz2v4VYYpb2M9lrUK5sfDuIRlubB+NsO0J2MS5dBkfF29XduA0RqboMY7nzw3E+7ivJF/rjvRWE99a3uTABt/bV0QKDJYAbbhj3XoRKxVBFw79pI5DgcuTxrzas2Lh8onPcwLb46OuYA8JCFFqzlS1/itxCajHeIdlgdwcPYBqhTh9aCsQMHbNxhFS7V2Qj0nic1CoSn8cicQIWiZdhkfj8RB7aCgvmshd4fNzPsSAedbQAd6U/TsPHB6dp8RBX4CH2Zz/gxUgnYCvxcAak4pSRvJ/Wby57ufViWygsV8Q+vYqJ/tlnXAveeJjsf8Zkv+Uan8xAx+NZ9aBvAvKZiUMvq63DKMuL7hG5JqVXEivv3/DyjlGXM0HrV56drELcpCmI6XR72UNHza0tVIFLrJuZyXAH+yHz558LtVNPZGK2pnrd2WUwguE+Y0ai1VbxNdk1x1s+/sTKLzvzlypwZ7bj1R3ER13CSSP5fVo/jER0KlI8sgweXg3hzfDA1rI2vA2PM6A5F5aUZ+CFaVYCg5rroBkMcpEBLAHT8ubBE2EkWoIYGvNbeflEt6rdrt9Fx30RHi3Xmi1t+SSyY5gWPKiLBLvoau5jP5Zr/lm7l5CEESKTA+bF2vbuigdPy0tBVs35QEuAsijl7lUwMRHcRjVvYJDBKruB/ZdfA1otbBzRvJVB6oIx0GmPOpB22eK42kbkgHQeFFxGCotluJT4hCsX1LjhJHIn/qDuGqYC16u5Otz2Z+1K/X/yG9a68/igbt9Zdn819xgDVSJj7W58Ht/g81iJz6MTMe0U1cAeUC9uCW+8JXryXNoo3olgkNa8gyIDPgUz4F/yiGB1lLDc3sLPF/meLWUkJ11XStZa6hVhU5H+1xqxPCYefVErPEFYt0iXImtNCRbWHWeP87CAdARTwk244ChWjlOsyQ7HKfa6lCCM3CURTUW1t3QqZRihHYeRmyVbkt+NPSXEEuxmyb4qUUTU7wTbdbXcDVOe8+zMcjd68FzKYOH2IT6C4Pwf8te6CQtLMdinsu21XA2GSj++hnLcbL9ygcmiZOfc07xvKwnXcE8uJyQlxKgFGfN20c0us223e61RWzInLiaeSFM9ubh5osaa+cmL6rPNeskAE67mv/rRck90nrRcjc6XVq5TbFwjXjX1R/A67qjlMsxwF3zMcE8zzHBXb/me+LduYw1MnMvGEjXMN5c7vnDflAVJiyu54vEldzy+Xy55fOewSSyYm7JbW1z/VkRf5qFpJLu5+dW3SW/JdfqtFpe/0wIC+BfPo/mCp/9QY88TPs8jlX6TtCyZj+fzoNcX3bZwjHDjKKC5kai4S/LHsYfhR7IqrRoU+HUk6938SqdAwS0UPs0cxK6CgfPYmfwrorE5BrY76rn/9MISe28YfIhPFZer1aMwmq8qy8Isf2oyu/PL+iTGxsdEaMLORQYFnTb4ctMxcOlNkO3RQHImDeNml6C2yGC8KRqrdmXQiNKh0FV9knbe7LTGRlxr9XitRlkatRTpi3cZ9OGkaw9Ucgu8p1xY7o/2xJeoK3sK3pJhWBDdF1HHHxF1QTg77K/uXVmcXgsaMR1iDpHkBS1JYvKL54qTTxdUT3iY+TSl2r6W+6NHJM99cP1f6fFfRv93evwPa/6lx/+jx/9Kj/8m6X/Q468R9OaxW1gdV59VC/NquBqhQT+ruWEaCwZghIWNO8gIb+EGU8xaD1nI6BT/5fhQf6P7v5+5iD/Cv8tT+N0/Wns4l/v4VmhUliTdLDygmUU77t22dZ3KZdu5uD0aS3pakk2Jmnvt0Nz4RddlVKI4C8OQvJb77aygVcKwEjQMC9pXWsaKEIRF+ckdc+yK/WXeGAbR+lZJTvd1B1XHnLxOviQe1GOmzcFYAN2waR1eYe51tnbtPBY30OZdr6LVhYqN2Rdrw7K5lxuF98psmisKO3j8VJCKK78Q4nxFlXwpvvCIZj69aa//jm0qrmiDe8wKFfdy1VqXybs0XFlN83uxMzobB6LCwyLRNqEughU1eAX0OFGDZ/NXHywkRWFaimWMVqMH00G8vfZcq91DBWaAc2qkedqoofzMv3jlSZwsSrQ4EXzEI9S2uQYnozmii+hGd8mdGum1QfyBVJKoxoSs6Bk0EPuJq9dY1K65K04masBfMHo4giVEsDJlkSPRPOr0gOlR7NvQJbwvWCBzd0WFuMBx6w7zDS3TlLU0VyCMYDBUsM5i4hdQ6Sr5z0gWrWuu+6aSKtwdUQ7D+U0wHTMJnVK/ZNd8Kd3aheVC3WoWWf6jEwcwTpjI7hWiyGEKY+EUuRooFLWPB3QI4DTYnqn7HhSC9hxG/L+3TMFZFd48WJHadouQWUQz23qBff6fF9gJN3humTvmrsz5IoE6wQeyLzHo7VouGFR4Rvq2rFnEkjSVLi0HD3BYc90itjceV5xoV1FwZRB/MJVMSqM9zxVUMlyjA8tViLYVV1kg1hUaU3Zei2tnXr5CZIpSiwQVpjiPgzBTtL2G+/M94Yk+YZ7oT6Jefnv4TvZbEzEshZpr+b8PTKHm/hRjU2Bx/oL3pR2XdxtgSvIJ80St7hq3CB4KbnEh3XHmE4ibKO6iwSA+MqEz/7XLxCDEJ5nvG1AE3nfeFEse1UiD4LoyF02niXe+6MSv/qcT/7g/7/z05htH/j6zGDkuL3QoLktfjDH+oxquCs6Sverd3PB/DAPAcK+/iQTAVeliAfR15pMSWDAaxN9LGsv/882Sb8r/VvF7U3gjkvV/K4SfCK+WsTiZII7M2gCwxYT8Bs72xNk3emNyv1ArOcSX6eiHvT1LbGAUpb613BPhZ+Gk0rR5OP3fh6Pgau7ciLgQT6IPpB+3pxw3rWyV/IhAJLq0fpXvCXOQeDkhWo37UUcYs8zak5igH6nlMvVg67/vzLh0X5sWiiitfckQfEW1tSSIz6xdfZZ1qOUizfYR6w2cF2mGB3DZbITwSbSNIGr7WKK2v0jU9pfNdvMryWyI9jhpEI439gp8ys7yXNXPrJxkqVqz5pDNgbq0rPo7k+jJwaamkhiiffqqJJ/eMouA1LSW/1K1aOqDy8HFSgtedP5R6Qx19mkjTv7DG+hXbYQFf4yFQabEj2YZ7s54Fr+OJKcWJ0wQE87wvz5XRL1NeBr4jhOijOFRMXI/RHPNeKqU1B1UaA6Oexjk1sxRnADfCzfnsfCLX8nKDfysEugznQWnei5RWK8zz+NCwHx01sA5l9nZKiSdtgBhLnn9fDYsNILnEls0jM+lMp97WSwGI2Z6icRML0lnpsclEkO9iWo5ZhF46FMyazrGVlXGM/i3xb2L6LzQGHa1argVMb9D/TCx68MycqHiSze6EnDoX0MMxqKytDa4Mz/jzgRm6cwFL7OWq1jOKWse6teHbZ4zg/+luG8GgyF3sGKfSn3u4z5haF2nrn61HcQgrb0SSvbwM0v/bmheLVfsK1w3rqExUxfIh7jytZtY7pHvv5oo9t2B5Y/be/AUr20tyr0io4L+JC7DSeRLc8UePPcIj4d7JTjBSGXtc1MaF0GLSqD7hm/L/eErHEaLWu/X8pAvLuCLi7711TWIa9zBLBBv1Q6Kt2q4gO5iLUV7Q1erEPJlakl9T0n3tSvpFTyZ0XffdvtpdXR0SWjonj1B6qbmbjrX/teXWMj+R3kyDgJP42nsj2vaxrZWxMihUZwusj3mY0goJBOV0Xr516EGT9XT+WxoWAQvLyRzswEvdpNobKmLV3ESbae5mq08XmjYDiPnsYVfp4OAfoILeCJeXtJNxHbc5la+eY6utG4OvL1a5+CTX/B53RTgbXslkoWj2q1KO1bXuNhrJnDdouOzVTCo2VU30FeX2MJIVlAbf81D0ulkJ+NC4pS09McDvBZAQixs42GzDd6ovkKgC89V6w0hzz1Y/DSAb56RSLzTfHey84m7mRxu7syBnzKhfc42XhhYMqeU+1U4dYnn3vXn7Ta6L9c4L6K5X0ewzWPgMi73507WTlfsNW5pU05qVlqmewgurIApuCyMwc3+eUS4j5v9024IecbN/mqHm3VHCRP45s7Nlotx3UeIY9qnI/EEIJ5Ov+hus8DRY5HL+hMn3TULGVzAJyRmx3WVHOpwK1uyr2Y9yHqfo0jLfl4KQ3ELnzsJGbiFv7KGkGfcwl9ZuIX5MbP4taiAdO4PcUyJ2XE5sOm2AszLBUMyshahCI/sr29G1oJHlgKXF/LvUDMpmCnOHzjeUUTBVjDGbf0inMdNvcQt/YIb+oTb2ZgYxqMfW4YR+Ku+NyEwB7g7YHhTkpCNT2JTKdDZUnjhq4Qp0P4foVqmoPatoVraIxVM+TZUy1uQQu9/hWrpCgplCXRMpyuvpVTXOiSN1CCGLkEzlMDFM1fCfa1s3D3tNfLwntB/esEytzA34SUoFZeefNbr6KuXE3bns56xr14td8gXfuhJXNz6+Z5opMDw6IMS4gykNz9teIyam+l7AVyTAy5tC/XrWMRsPbUh0P7IscIf5tI7Z1LIYO+MqZ1NutyzK9moPrNrA9qz7LBD0OrzHc0ZTu170fPytlu7DLgffdcJfsrcmIS8I+otyI5aEOYZkWOCmcc85XQXB0s/dTi4Uje3nFs/30R+z68YBhZVFsHQEklFnRTjW2XW+IXmd2hkeq7WEmSq7NSg81GaQsbLea3XAtWg5emgt10DE4rQ5EqYXIQm0qA8detSlSo9Yfu2RAwXvH5F4BIVovtaI3MPzTxaHry/BNYXwdZiyS810jhIUeZOoQd7rlzkqLYEoQSdP8ikxgcn1JvkwU9F6Kc/mDepMUWJsTt3XFDfYwLdV522UfUxtUZdNF0wB5O8H+aZFxUXKcCshksCM4ItXPGaKoqQohh/d2GgE5RSo+hJaBy1YYbbPNF8fmv2+nz1BBjnyUNnVEp1MYfvpqLv8Dct/8O3SNhcJImrh+OYy9+Dt4i8sQ7sNZcZWGL6DCmR0rQvslavZpDt0+FYPjtYpBy48jH8qLnPgMnl8ucNVyehTmpLBvErJg5Wy7uGukmAwzVxGEe9daTRwGmiN9gZSn5if4mwskQS3gQ96qThxvCsBPkeohtfUW+ablb8YgKSuYVIamnr5Y7nhTkTc/n0TVVpwmozs7ULTQdocEd+puShqE0xhJXAziLFH9A5pHZGHWbPhqA2Su4jdKgRvcGsbk08S7zBXheEJN1sNAHKMhdRVot9tzqqZzCBUXGnM1XlyY6DTZeumkK8waZk0U/THAZ3n2s/zMbu7FlHjQUu6h6X6lWgkr+oKIFs6CkJegqzq6WQ3F1ZDeNfl/QTFk0m7mC21CziDja7hLiDzYBM6jDKpuUhqE0JXC+GZWWSjHIYUScV+wd85Sv4AX7GXCS3yW3vnm2aQHCgwLsMZdHu6E/q2MXkI1dUv6Y69PzZdmGPRStOBa/FPfGIvbkNi/KlvsVwqwTWFEvin4L7UynswKvUrvA9fA/GlhWImmOzeaOzBlwPP73yUnX5qp9/qCaPOe22/KSVCpkgeqjO4WviXSSDbuoDxcrhKyuhJ17gs+HJIVmqh0nLu/dePXuKRn5ICClVhEHTqGou60F35TOaCwUXLUMVoqbJELybEV0NKooV8U/mPIVFNdOfco3gjRkIk8LG928X3piQSBzB4m/m1pr8Oe0BRhfSKeN7Drm25M4qNdc4eeUii0EmqM3DKdBOXaI/ccmVh2Vpt27Gav60SVoz3sTCbul4NVcVDnXKodecE/NN8hOT7pdG282atXH5wpVqeRmS4gkogfRiSZRuDiYgqRJ6Nb2CC6In2MyPqDP6zqI7Mkbf358IhlkZ4RFRarR695TFXVUOS0+ecNPMFP3AslSZ8EE5ZtWV35sS7pQ+TZ6Ksdpy4gfmvhjj9oa9PJe5nOXqVuFfD/ybxnKZzqxcq9rLXyUBaTBjjLMn6pF8PQ/yRCD0nFkuK0s7H/1H+fxGyXvNCWYV3+nGnEZ1tr7dlqhHmgtMZXhkSsqF9RPUGxkzt/V2avkzCKuQgPVDKViDwT5efM2B/0hzIOyLJ/MvwIBWkkdcmfWBWcrWg1Gr6cjsVwQugH0CE4kDWg58wO8vTvHpdcJgIq0LTN0u3Quk10pz6/z421jktimB70okYQIvLTDOpyPAhvqF3oBsqDz6In5+XDKFPgvfUU9KJtMXYBo1g96MplF9W5834eexU4ppT/QdNWZKCV1RoiTAv4gJcgeIuA2VWbtYiPBkxdBWrx34fmJoKzThcxvHvz3MG4o4uwbRuTx0JyvvBbHZUHBbAlPKpRcgdiEv7w7hhEwSR3Ku4hsX8pSvHuRphE5XYTJdMYKVTxLaZsA59zgeYqfw3SEiGwZiqiy5iqmyFAwhYhYvnwwRWTA8c4M79MpRiHGNsOQwAiKU9VfvPdVgznlV+rhV6329PdSgf+jmufuqyMiA3ac1mcxxn/VHV6r6TFiEhmnWkHBtelcnNan3ZSkXu1+rxgVDmOOBYadiValhGxdg2Wihm7ODWt5Z2/5OKu5NWGvDi9yhuy681AcYCBFTMPeGyfYHTLabJvMYIudjJu6cJKwYjEswudXOVg61mzNi3GXrxwmRp4JCNKhNa4S/Y2KEvx2tEf6yYBiJ8GewNLe/etGXCH/+Pm6+JMLfhWxdhD+H3dlQgRdltyebjdckXVyTZeKa4N3wS5WkoUjaYCxGTurqQGXzvfafYo0H891FRgXP/2toi1mT375hTV6TKf8SOUBBIg5xe2CycSHNmRPf/ehW132rNTyXZs523+XH7hrBT97lw+8awHbGbaKcVN54Izs5wIkP6IMzvFlh1EDeAX13Gypu72LRd7N5NAEZCcPGVpGdEko8JIpozg53kdF1ERmdYnEi7iXqcAc6Zem6iTq0BiaLsTmhwXvktyunL0bdMSnQRVf0I3EMj3+Nrmi70QOPZiEejY7NWyiyecdFNm97SIwf5vEmuZEBxvHICO8q9J0f7hkeB/rOh0ff4ZGAPx4H7gUeCVI48UjRh2R6s5CDBwOlusFAKRmMtneVg/t9cd+7NsDDgzweBncEDwQenmLhIRkIFP8931C8+L+ZccG0dZM/JJu81A/XjbsDpT48lOLuaMfg7uC6cHegxImHkj4k05vVbhvIz93D3oS90puHefjuq11a2h/StMO83JZPxNvwCmbEhQtTsADRhZWj/v7F2mEkloUU+viv4235pnwM9A4DPRWB3nVhdwtnyYWGLZGmSd6f4g3OX4N38n/D4FxoAxqSDVcOsZYt6f/OZZ8rYuoqn96t4d7jyuvErPdd2KVC0jy2tRfkllHXkWH8YG3iXDFdCIVeX1OfCI0rSaed+YQv3a4dxN9NWMcjE/9iOCbeMd6uTb3Grm/ugxN76kb3oFZ6VKCU45icBAptEkNUxoYcPxGkzmb2bN/2JX4lhDK1WdQjxnQhBbGm6CkWZV75F6/Y8I1gl6jt6J/BknujPdu/xr0U/tSJd+7fyonrd7DN97DojtvKUcs/6OoZgiW0dKGfvzIGt66rhbTu8qUW4j7HebuyYhXVUdEl5HaAVJMuRtXUmUJH5YBUZPXjs0Fd8qYM2k7ikUTk03texrx5T6qcNoMh1GEm8eTFuNsmjx3iR6oRjXNGK0GaxCSf97V12Oq1SiNf7ycazEZ+MZi18yP2sqFuwutWi9mScuOv9rL1or1sb8ghcQhmSDMhRwkzytAMWr4T7hdKkiFTmgz3v1pReoiJQqyY2GpFibo9/2pU+Fw0Knz+xajw+RxefrIev8VDlRQS6sVghta/gBU1ju6PzCknZjqahhP7gfVrMXGAmCg/qasBl1nwHGeToIqXmAFgjiFeI2sKHUJWrbEWdcm6WIvyfgO0byQ3nklh1YClfF/yAgPrpScHLGF7D/jq8jkgnG99Ex07BwSzFv2++Hed1fl3XetnhhND2X7j+cHztBcl5+CM9MU8JczRXqQm0Gh2y0VKPohknCV2ZTHzlGi2mANzSA76kWS9KCVFSrUXJ+Npy/LLfp8zPxdC7oBBjuJy/tMqqMjn6uClnzIvPzgkWXOL2bbGcuskFRowqh7Gg20T/AQToOf9vqgr4laPnLE+ICTqsOY6xFBT6V1u1G76jPeGQCdVz2l9UX8Nlzn4sk1RStyF6FD1IZqrW3MiwuOW6n1lA0yD0UNAjmYge/Q9GoG2IRfQ7wmzcYmNsE7puDX8zvOQdxWRob5+JzROyIny8tri7aKSozy57ynh5HEwOxl7jEbWR5lStlZWdOigkVHtwSNGbWFBB6GDMonTCyeRyaboBerdlhhLrraRtOnQZnEbnzZn2oS0SWvzUNpGaiAdI90rPS+9IL0izZUWSV9L/9L/Xr+z/lB9S/1F+m763vo79Uv03+h/0AdKRs2krCkHagPlQ+2iAqkwKoZKoq5ReVQz9ZmmaAVtTJvQ/enh9Gx6Ab2M3kNfpnPpSrqW/ouRiuGsezH9mOHMeGYWs4HxZIKZdOYBU8I8ZGqYJuazgaGB3EBpMNhgjsEagwCD4wYXDKIMYg2SDdIM7huUGDw2aDAAQ4mhviFr+L0hb/izYR/DwYbjDCcbWhjONrQzdDBca+hs6G7obehvuM/wkOFxwwuGsYbJhmmGDwwLDR8Z1hs2Gn5kJSzFdmC7sIPYcaw5O4ddzC5nHVkPNoA9wJ5hz7ERbBKbzuayFewjto59wb5lP8kkMkZZlPltXNCBS2+A0f+kN3QV7L9RFTIHr6TmknjQPe82Ym5JtHrQ/L9YPbglZHqUq2THV1NFQlv6/1/dLNvpREIQuK0Jm6NC+uPHo4kaJwZZNCx4nZN0PiJMfXgh3X2xy8D5C4JObsbi29aYnG1FqnCIVpotuwrGJIyrIjntSUqSt9d5NSpgtno6eiwj4/kaOZH7YPe/U1Ue+W9VlbP/raps/79TVYb+D6pK2dm8kxmic/qpRUEz1QXonC4SwElqDD0T5VC+s3foHNP9k70fqCdBNmVGT0bZlM8cPztd+tXt99QzcPowWtbqD55OPAuJY2EPXiZtRNbK6oQbr6M0MF4ookwxbnxC+TntdBKvAnaF+aSqe8JjahANo5qPe9lQMzw2uy9XeXscPLhdM4fxD7oQEKqS2W12WrvIZGq6/TPC8dq5hqUkXwyPOqu5uzHK285k5Xr35WrZta7/tvfbSez9svabyGC3fydWdsr4CZ1JzAXXi+aCz2hko51GAk7mMg9Co29EBO3Ze4rELvXZcmKNasxC666aqSR2KSwyrodgGmYI43Hv+yM3ymmM9yqdn0uKc6W6L7iR3s9qHk/JIMq3Ky+rg5tNJAjzTeoljfpAPnWEKY64dusPE9BY3ENd1GgUXYdqlXCyinl1aXW/wWschmnkRACorVp1C3P/27d5aP7m/ne1cv/HdNz/cOLQILk6GfP+mTre/2/Of9N8wvZr5PD7fzG7o2X10l6sLIK9V5E7ActYQ0BFTaRliaExl26a5NpdHTV62ZLZiy6uS9yiloGH/0+87J/msDbNM6k32mljm+fTMmIc+cUwkhhFaohRpOy/3lLqrihlQt1C9jXwJCq3dpVyBm050bKcsqCfIxXmBWSCcTm6eoiGDCGTyms2nAiRexiU3pxFybbzMNP3J1bWemFcMLI5l1if2nl5HTq4vdX6NAafr5qGkbRMEl0HSwdh4EJm0tF1px6ZQC/m9vZbo4LVh5jq4MCMLJJQFXOy7Ki6kEF9drnMGGaC+jDTN2+YtFfNCVMZ7nXZ3vCN001Qb2Zo9AzoswszNDqNyn/Vp8hd+RpBwzTI1rNDWVnUBccUVWpCYiEJ5L1xt58vCeSdlXbizCVNEePntH6HjcrcNrzybtDFK8ka7vX9e5EXykzujrFmFnt6LV2mnm6xae0oE6KCr7D6Q+0ofFC+bBpAy5W1Wdcq7qy7uPSCRqb8okKT9cpQHmJ+hQHUQTSYlu9l7ZJHzF3stmG1eresWFFVC/1qZKloQQEMe1SZB8vy52fPz5MV7G+1yZSJt+KrWNlxe6p0OS+L8M9fs4SXVTHQrnwOmq2xYtDkOWaIwyLZPKCUTs5rPO1UVi6Rd7KjIu4d0zw5SS06bhu0Wo3l1d8x3rbpTND2b4eTM++r5EN5uWjzK593+G8GURa6GK9ICS0TteL2PBdcyXAFDiwXRxTAJISrUFoq025RFtIZwGIMgkYjFr9kii8yN+00d1m+VQGw+fMKZcpSuvppLfSE6SbhDJibQt9RderJ9OO+JWg4GmyCT8XAKajvmL7kfw++A2NQIOP0dKSA76E9KG1uonaIQ3LgrBZDe6TEz99TJM7xPbwxcbuTEI/bJS8kvLTEOrf/AOupndVejKxVhyDDg76pbLzeep2C9G7MbVQLKpTzJU3mJpx0lx0M8D3oq1pi5+K0UbOblokOTjr3JpkyCnphxm8H6kXJVHZ4Xgrou8B78rLNK+1tNAMRQ8nqtJIvhnO0HFOpCJFKRWIqNZfp4kP18VlkM9ikV47Ns4obMckJ+KC+3w/cT3iOzdzX26mbf8Rcb+sFT4V4weP69YIHUyGjlMr3GiF8eHM4Y75l7Qx7x2PHN6qnMj7RsTtvqOStdjUwqdq+VhbNg2edVCaqZmTazVi+kolhg2kqDW2dR8IGy8n2J5vPWtx8sD+Wl4k6jkklc0plVaF4W0UGnbsYcVa22Ym66GR/xlql+/cJjSMz/dSCCw7quaNIvH7xWqp5kJuyjIZm7WBqEp3d/Epp6yYOQtZEo03wFzhBHTWARuvQS+onBLtZ5eusIl0kc01rJHOeRDL/mUQylx3hdXboxAp9HpvzBwkh/fVCQ8as4GVrWLnkQR2crpbKRKwkvw2DMn/L/nJNIxN1aGERvKw1zkO6rKs9Zc+UnaDwLGorkJaGH9BG5au3/WmYpO/OSyLKoC8uOV6mPMEKx8BCIjvK/v03CzKY5j8VLzfYFSM7TCMLqiCySvbCjxh/yrSeytn0MqShntMyTGxcWdltYk0iIzYeMhJVkf87qqJsHi8XTYHlg1k5Camou6iS9Q6NY2WXfYn9piz88InAcyrZ/RBcEfS9jVtdqH2DFk4HumUNlrLkx3hYSuKN6x+9En/XRLa/XLheLiuE+mKoK1SA8ZPTNdxvUbJyb16mfE4HgQVljg9vEX0SLPBGVebjsyF6UFmNQ1aox2ToIfsScgVjHqkshZzqTxjdtK3IIQ56U4tlMZkXzu/ecxrT1WPbNp6QQYp/vkUpTCiAXqXmRTISdfwi9MaUGmMzN2H2o53u4e4yct2gdl5Ey4h5BDGukGHhHJxqxJDRJBqMzGHrVhe863zpwHVWx2WYw9B5v2Qwe2WwGi3fycsyiV3B3h2qDZuPxcgC+EIZ8X/hJbJqKUyFImX1s1G0jNyCXtHdgsqeeD9Zk6su/p0aT8toF+RKFdLR4ErJVrNyN7hZAifLZeu2xNlhzE5kcxlM1pmVZ7jLxBA260gIG5kYb1+u+59WVm+npJ9em0lTZszXkzsv3+yi96OevviPm5/Fv2Vl9DrrUeMXT+6k12fi3Pmd9IbNnmHRSc9sruXMTnrT58+1IH/frIMl/52pe2qjZ9j6JMX165709WStT5SeUesTrddWT+7guslVb6H4vUT8the/14jfLuL3ZvHb02nlRhc9X/F7l/h9QPw+Jn6fFr/Pi98Rrf9B+7/9lvyvvkl/5a1vCnFu2uh16OCjJ1m+B4+JwiOV2CvEubsm8ZCES5IkNySv28jb9GkzqM2kNtPbeLTJaPNRqpD+IB0rnSHdJU2RftDvoD9XP0n/HfUDpaYGUNOpA9QxKgWLYFV0L3oIPZ22ovcxPbCYtZN5wggGYw2cDfYZHDGINIgTBal7Bo0GHwwEg78MOxh2Nlxh6GjoaxhuGG2Yalhi+JIdwM7CYtAxNpgNZ6PZVLaR/UtmJDOW2clWyTbLPGV7ZKdkwbJwWY2sQfabEW30o9EgI3OjZUbORp5Ge4xOGSUZ3TC6Y1RkVGX00qi5LdvWqu2qts5tN7b1aFsk7yr3lJfI37UzaternXk7t3YR7bLb1bd71665fZv2Bu3l7Tu079zevP3C9kvau7T3bb+r/ZH2qe1z2zd8Z/Tdsu92fpes6KrYqAhWRCqyFGWK3xR/cfqcnOvAmXBduX7cEG46N5tbyNlxjpw3t48L5hK4FK6Mq+Jeck3cB07g/uqg30HeoUMHkw4DOizuYIdnXvn5k54J/lh+fqXniNfC6TOSuOtREg/82Yo/nvjzEedJ9aZ+LtKb/rlBb8bncrx+a/Xa6Tnhj0yvzecHep6fa/W8cL43/mz7XKW3W6+dZOTnIsko/JmNn+fhzwJcuyGup4NeR/zE489g/Jn2+Xc9C1ya/I9uEG7rE85n9AJwz3bh2kk9s/VYyRycM0+vg1iHRK/bZ4R3S1fc6+74vRd+64OfJXq98ZtMrwt+74r72etzk5g+Dfdn7ecmyWk8sjOfGyQh+DkUP4fhvO900Ljcz/jTDbfaHZfqiXN64X70FmtGetNwugXuxenPnyRncU+C8e85/BuCP7gmfBa7YohuuIbueCS98G9v0i7+tcCjwzMlOYNLB38OlpzD7epKfcKjNMA9W4pnwxn/Xv9chUfXUeKFf73xOz41eAyGej+Ko0F6P/9/7V0JmFTFtT7ndvcwNDjNDqPsgzCyyaKAbOISlMUlRBQRcQFxI6MfqJ9Rn889hkSDG2rEXVFBfajgMqCgzmdijPP8xOhoMsbMZ2xjWmIrdnz2e+n3n1N1b9/b0z3TA8woee/WV6furfXUqXNOnapb3RclBgte2qs03YKaV2eSvB1xIWieCGLSvB1hGZUjXcpJ3wxWKfQprZTaH/erEW5F7sfh16MvG6Qk4jugVIr62NZMiZRtLaWtrUfOp23uGPVAr6WdPuCPQcBnX9wPQbwplUSperoV7a+SsVV846ghiRqS0HiCcx/cDbKYmRJJ5ExLW6inD+oxo5nQuqUvlR5WCeROILdgltL+CGamL2nqCSrE0UY3cFo3O8pR3yjXm/7hfn/kG4X2xqDFsYID4m4VPBCuRtxWYP04xmwDxnAjvNA4ZvudsmMjOCWVUvvDj/HGJ4HSSUtnl3JppVrKN6qgmtbgH1lDY1bKCK/HkTeJvEnE1EOfDgLcF7GDldPrLK+mqJPNm7J541kuVkmLq4TfiT79CnJwF0IZ13vhjUyk0ebD6GEpuKpceWEG/Cz4s+HP1bFJQ0pFb/8U/j5Q5liVTiPhIciupiAmKrIO/HpbeTcUblDJgESitrjmvk5LdKM7gTV0AKgeUaqbesulXv5IWkUPHFu3tH8tMISGIJHBh5F+rLbv11Md8RwBDknlgjGic7wepazO6SujbOoXWdFvkveG74enAcg1ECVH4H4k/AF4ngIv9IBeo/nwJ6Pepch/Pu6XIW05/Dr4J+CfhH8BaTUIX4MXLvsaYQr+H/DfQPo6wc9En2bBz4Y/Cn4ujeFLwWuXKb9F+Qtg+CXqZ+kndacS9O0S+EvhL4MXnXs1+n2NUjRmdWc5rcjUQLNsC+hio0v7oo0oL8nU8zmqcWpRazl4SnqeHat61X1m5BOoPREYLyMd9b5aRUPHeIFLT9C9HUql3RK+nIZbOhtdJW2qBEbRJsYFoye8dSfSRG+Y8RUZr7U8EUEbLk/EWzgHNTe7LNL5qRRtLoV/AW1cCn+ZSsZCGQOeg74zpFK4bSBygwNkxFCz4bSYcpnhrpjlZtHS0tdfGd62UmZgMZLUCTosqTOV0RdboL+SkG3ROinMTHHMMRHINGYoYLs6cwPmmjjfo3KdhFwnRJdoW0ZDIw/y6yyG+kqtPsMYKP3rMd6QRZSCzrb4Dm6EUwm1s9I0G74adR4IPw73EdW4qrNRj6uBulhtHrczkqufkqr7Z2DUwWl2no5jpo3amTaq/bgX96YvaVB6kM6FSavZUiiRUi0mqTG0E7c6PoFel9s2pO560CgJGiVQ9zrQKIG6E3ZOltJJjK9QxEGNCbEVAiMV7H/L+K6En0WvqjNbeBMw2QxL5CXgvEU5Q+alSrFlUKsDDNPAMAV6DUJ9RssnrZZ3aVUPHVQLeqVEiwLXOpSqt/q83tgZiHPplQ/7fHH5chhYphIxX7l9lkqGxEI6oNNmQX+VQ3+VQ3+VQ3+Jxp7lk5rG7eTn+NaKbaqPTT2VoK8RNwb91FitMTsLNX0XQg0xlBQ8YkXNWF0hU9DFGOu94fvAS6qMt8lRq5Y6xhqS2UAbUGIj4p5F+BzCavDcgfDj4Btb7Wb28MPWmkn+r9fa9KzXeqkt00i7c95sOlXkoAJyUIGnCn0y8qQzLubTaB4oufoil1C3r6yEULdYJcfAi+11vs6t5a5txDvwHCxj5uuwzg136fxgLNwwy0rzLl3BRY1+lBkAmlfmTpHN9mqh3wIOEOtYyruzxr3QsPdZOz6say6jrdM6a5ToHL9VZ2DT3mMI11PEWyml7OpAZ1bUnLaYJWQ9itxJWbFgRG7BaJg1U73awndq+ykWW9VodymR4Ee1VAOvRdw69MasNWSNZdZ1G7U3sqbYilx32XnfUCKJ0oJhUkqDG/ponxPeekraNBZF3LdCSGnu3nZ9FVF7x1iLcd/aKabWofRnrVIkrqtnWQcrvXQVvU17co9dE69B+ChaeUzXWzFQLcpPI7TrLmhGd3VqV0i6tjLWQtrimczWaEZSe7nGjsNauw4zazBj7aR9o+yOpTseJqfDq7W+FGZws25Kaksmtl75JexygLeKaoe89XZOrufVSsFaWffrvkPI4xpZ6TJtAg4ObTL4gRvX69rPjU0qB8ldXMfTpEQAN9rxTrjjbdME1gPz55H6ovah2t6FNU6s2mqUkJiQF2PSn8OdxFRjTIIl8AR7RfB7UXv4rM1paktYuyYFu8bNE6zbb0+wrgDde7E2ZU8pSuf6chjbeF/w0WB4WSNXIgesPcAKWUHr3kba7tdEXDvTcpeRZLOP4tg1H2hyTJnsNB5b+qNv6RxoCxJeyKQzsiKLqOUKmHX0Pb4yGzWISB8y9ZmGTFLxJ4QJPG/LQA+gb3XoUX0mjtgkoOlXA+62IDWCnHWIr29jzKuh5QzuUaG80jmqdqHgX5tZr+OQ0rxpwVPibdl0Zo3KL6kMoBzS2/LCKhkuDYtA8bexBvM4tLZyk8bF1OYjD/cIxiIm8doj6Usy26/WvrBiNDQWLOsx6nbMlf5xPL8ucoXYiKRhDIRLkgZLPG9ESky4Bc8JlZG6tsLc1wfQXbBXLkgLl2s0uFjlWHIIP6eUsqaEpNR5PTX9qWlTnLeBJ3Ljkh5+6E+LamsjjZRZTR7fBrSD8DQoCAklT0ITGp8tmwzkF35PtC62/gvaPib8rfc51BJdotyTKrKudFvqFuXVZKEW0aealox/G2tF02bK1dnN0c7qfHOfzJdXx+p7PgMXvtoO7yzF89Pc0ND1XmzBsbGatZW5pynuyMXVuypaEaEWXHsqT+Zenp3SfM5k/l7rTLuHUsPFu/V1TJDTC7WWOxZNS2DbzEyefRozVmORpRI5NoMbHzd+T7uC+qjVucVnLcKWbamVHVhP2FlB1ittMJfC8spKVVxdgAsKzVHArUHDIrVR61yB2TEesFhz12gxX8z3YU7a2TVktPksbXjF7J5A4/j8V/7c38UVLUBJjc0jdd8Z3QM8XvzsX1fYJvtuZXbnrzac/X06UHbmiizVxDzZlqtpr82C69JGOQtZ73vw1YarucAarSU0L5jWFvsAMauly1tQZnrroFL8JZZWpjbv7lxKd9TXqSVQ1Ni7e4ttd3k7uL7ZP6vNgE/CjfPxx/RCvNJmejzWJF/rfnrR0lZo3m3lSylaX1BDx3SHsT6owzPrzFoqx56Pao+/P5bMTl2t/6bMXccUzFCA5jYM0lw0VUTfLyVVbxWyL3fL5aeMvIMrOmdefZJJ/P+OaMtaanoODa75msoHyreBhvTjujt253zzwR7GNd+RfR4vxAWN9yyaqDHeNrOpmeENnxSmU149ko9bPH7z3sjvMVcbcksi217LaF6wxrZ6o5s0lq0+mFMJEf87OJ1dUsETCXje4p4pabuTCrmXzobpQlrO5knl7DUm2mjl0+SVs99S7GquvpBO/1eY/VtdQv0WYPEraPc0TOE9roj/FNLuvjLb4OrMORbl9hpIXsLDyrdv1GiGiiK1xoenrAzz6fw61CmrWql5d5/3yqGLb/Xpt2gacvPa1IgLm55VyZ702b2XObOgZ+dIdaBd8WclUHoT6EdK1w5mHZjI5s5be9odMT0zFt0tHBTVs3xBKzdBRctYMRKYnQl8Le221baeVWyQk1q6dqzPbHPjPZrLiTRff/TdoT235b6dydTmp7rMBeByWStt852z2hXdKbsSr8NZyVJq12Sq5Zycrb/ed55uW2Bs5FRdnVmLWlpGjKQ3wrtWJbQO4RY9f5edq3f6At2iWZ62cWYmMidFGwrMKmbszXhEs7tKBVrx4n0jYnc1dkVis5aoe4ol/3vMRrO/Smhw7ROUGC9n2sVxt9q8MR+/RbwzuekADhEfzF6RFu2fRyxnmzNAKbv/utMcoxJmzojWKw+6JydfB2fWgq+36NlPOYe2BjHrQLM6kVO415G2LrMFqSIn8czritmWvM1EQZ86lZyYaVF3Yutcyd+ZHRnf/Nho3ax099nt2XjvztX5LV5H7DrPKB3MvmdST+b6NVZCz7PCAockRvSpnrz1pc5LtSrhKZXUpJ7Lzb+CjcppTTuHplWmk1mqmfiW0j1Tbeq13CIc4uruBrVnahS3GrRYq/yR1JPFSeWamsxG8EhErZ6Uy0MFMCerZ2Pa825KgxqzlpFTybRzZ8BjRnr8ku9bA6e906/JIEXNOspLTSllC9HcPd+r+iBIX9vmTs6pASs37p7rb7ZUkpraRW2MnZWM3bXu0BlUuMZ3aiJwFrrBRvopNd93n+d0S04Lvr3eJnkipjNE3Eic9i9iR6nQ5Z2EaiI1eB3SRPvF1ZD38voVy+WovHZ/XK3XLJ9FvZWM/7cJ+fgxO0MZX5hXXdpEcsrsyhXBHNP43VzKp+nzWRYtajXALc3YKWqZ1fnsPG/GyJlVavX8eX02T6AOo8Hz2iJ520wU4PNGGsdng+XYQtZqqvd+zRF38+tas8HrEWZencNzMDerDfnFULHag5p4bxo435YM4hsYD1056Txi39G6mBqdTeZ3LGlbZwzzS/75Q1e++ShWIHdhmuc7i+fOA7v6m5JZnnRV0Bi4xunmMhI/hiq9lLkKm3uHaHTRZNWmMa+2Clvf3BaeY+ubvW2831LUTGTWxUXoidy5eTda6ZXySyxbX6XSIpqtHa2NUFvF1alRO5+LlVNpfyvYXP2CbYX+rwl5Y+qeCqtsoYYM7psE96ICT66l4duzHRPIWdGs1eGe5vDr82y4k6uM5k6z7vrZYM9+KCZv2p5RDVp36TwriFj213IF6mrR+568uqWpna/cd09pa4M2U7fuPNQ21kx2h9GsqIvV54V33v06OxXEN3dE9ZeE8ktVe97X6v+0tUyz1neSyu0KMV+LqZbs7jfxlqaJvuemmVnT7Y9vvGN5+IXcPjW1K0jeuedsGy3k4HiQCs3t0hRdb9qdY611nOcsdON3nGrzBN4pFcZarK0mTtC0ENcsBoHZogUn3ovDxNuJirdoj6aYmnPeQdvfozdXKlnAJsmb08ib+ZX77tlFh624zvcUl2d5u0Bm5dvg7Y0mfRSu8LBZZ0u5Mp9nDOwv3M19/r0k24J9s1EjusVY37oLXK8aJ/A+RH6D7dqKObuiabV0aylwUqepcbb70vn0uX9E3TcEMQ+DYF6ze+PutQXGxq6Jv7N3r01ddk9NcPft39vxTmWffSX8tmLz0hnZ2V9X7OY3v9YKCswJ5n8iajKrdS8t3iid1N4xPJ4idz+9UI/96/igJdfo8naVs79xN/JUmzMb3OzHJKeOJCRgo+7l5qzADL7Z944Wn8jOzCk5a8ib/acN7HyfVAndYleYria43e0h7nPWdBZzz/5xf5vgYW5tTOXNqLtSbDm+vthE5nr/+w37jjah77A2mjeDefgt5ntrFg2+98v/vxa2rzFNb07ezWqvL8JGOXdtXgr8hitdaB7KdzJBg2ZnfzevexJl109K+TnE1UN6n7S6vyZr5+bY+vHszofsShZre1nOi7v3AT6P64kZuSJ2PknZf60Q3mwwJ8vz1OnSoaCe8M7uuJcrk7Ei9Kjdz27tsxffzQXKyL9VxPQNRsTIlZ3NUxj9lL4/yd2Ja6Pfoe7+M0hmds3OAhSQOt0XS1ndki2TViko7hxXzNoqgf/cyda+i5gXPoGWoF09gWZ0Zzf/7zM9zHdpHJr7LbaBOfvATZ5Xy3u10q8Z9iypb/Z370bvFliBuTp8j+pz38BT29v6DvWyX2sIwbFKvkPt4ELUHk6+QNGB5N+au1AJ9UDuUuDcV/5rn4ZTR9qfRiP2ADqKymkZXUxj6Ta4iXQv3U+T6AlaT1PpebhDqRruMNoMdzi9Sm/TD2g7fUFz6WuupHk8lA+iX/DH/DWtRvuT9NsXlAePHminhPaBk/+q7oO4fnAdaSCw2Uux6QUcxiF9ArDoh9YPRtphaG0QHQE3hGYA10o6Bm4EnQA3kk6k+Sh5OtxoWgw3hs6GG0vnwh2AXl1IB9ItcAehZ6tQ651wk4Hnapqi/ZxKa2gtHYLePknT0eP1dCT6uhktvQQ3m7bCHUU19Bs6mr6mf9KPQPBSOpnLuIyWcGfuRmdyPx5A5/IQHkpVPJrH0zI+CBT5CU/myXQJz+Cj6FKew3PoCq7iKrqSL+AL6Cq+kq+kq/lWvp+u4TW8hm7mR3kL3cLv8Xu0jv/IDfQ4aPoxPcOf8+e0gb/ir2gjpzhFz4LC+2G8O4LKneE6Une4vUDdfagMdB2JuLGgUQUdR8fTNJpHJ4GKJ9NC0PBUqkLfltHViL2ebqJzlCIX0F1wF4Ii99FF9BA9Am54jNbRZaBLNf07vQha3ABKvEK/BC220c0Y/yQ9QF+BFms4Cips5h58DP2a5/LxzDwPLsQn8dUc5mv5Oh7B18Ptzyv45zyKb4Ebw3fw/TyWH+KHeQo/AncwP84beRq/wwmeyX+HW8RJuMXo+Q4+Q3rOZwq/gwKT0K+J4JYpdArG7gzw2WHg0iWeZMyz4Uk2LMFdf+XJeeCaGfCnglP6gIuGQjZi4Cz5ysDeymXuVQrOIXDjJBqPVifQmXBngY8m6t0k5fTJwEC4NAwqR4DJoeDVw70a5gIr+bXkbHDQ0fq+52z6kcbPBf6HgXMXAv/TaBGNAueORj+WoLy5DoQbh5alpYO1hOumoz7jsu0c77lF1gWv7vBH2vvxqNFcs5F/IPiGwCs/BE1Mq9Lf2aDRDFoAnhE/D5iZ91CSj2iAyp7kHq9+vFLEUGWS4isuF4firklK5bOVVmcrPgdqG+M8N0lbNG6CdeRhkB0lcZSD2SQds8k6bq6bat3BPjfNukN0TMW51J8N+rtuljcSszG+xs217/XMZcZ7lh0l12VHa6G6Uzx3mnWLfG6xdWcof4g7DPx0WICf3Dv/daR3dwB06f7Qsd2gF/ajwdAXvTX+RDhSb66xmBtG0jDqSp2gZfeFft5np8bQvUQrz4cOmm/dcZgBjpMvfAAeB/wqoK1mqjuW5qiX6wfWn6qS1N5KvKPzwSGQ1ApIWxlSest7QGA7DJIjPCnXAvCzXMNBH7k7HLwzAJLfC34QKDQLvf0h6H4SOPko8PVMYNTDYjsZ494JYRf7bMLOXm+6WD8E2mYoaNoVXBHW97kRlS+5ugGrfuhhP/SxHe57Q8cIlFm3H3xP+ApQdjB4cy/wXpndQZLyneC62JbkKzZZ189zva3r64sbbF3wmgapKIEcyMVen6QuuY6AdttbNaJptQvie8F1AG+IH0Bl5ptYmk90hqFqJy3hv7qoG6ouaA8VezkK+yhteis+4tjW3MXD3d9el5w6OlPjK7f8EM+FrBvqc12tC+uYistH/56++3bW/cAblex4u6M0XUfeP1rmucJz+1o32Of2sq5M+UNceUFn+H8sZrMRGG/3Og4aaT7kaQak7FjVSidAA+yPeWYUUkf5qDQMJUV+xsKP2Knx818j4XpiJu1pXdS6SYqrwW8/de2hW9p7+qW7+kFqQ48j+502HlfySuA7bQd432kT+W8H2B1UrkC/xkGWDod0z4F8n+p9v+1w/X7bWfr9tl/o99vW6/fb3rF1yPfb5GtIfdDyKPD4VIzlbGiJ+dDAkiOm33XrpJy5L3hkNGTqYIzpUaDvSdBtJo988a0zZEckUah5EGTvCKX5ApkDNY9YwF1Agf7gvuGg9URosiOhg47H7LrY5pGvxHWFtA0A741AXydBv8zA6J2AGeIMm0e+H9cNVBsICo6EbpsMPTcTGm0e5o4ldOai05Ze4CxVuEzhxQovV3iNwhUKVy46bfkZziqFqxU+oPBRhU8q3KCwenHVeT92tip8TeHvFL6tsE7hhwo/XrLstEXOZwJDUYVDFc5UeI7CFQrXLj37zNNCv1X4lsJ3Ff5RYYPCTxVuX1p14Y9DOxR+KzBMCksUdlTYRWGvpectWhruq3CQwqEKRykchyzLwpMVHqrwSIVHKzxO4XyFp54ntS1RuFThMoUXK7xc4TUKV5y3bHFVeKXCVQpXK3xA4aMKn1S4YTloHq5WuFXhawp/p/BthXUKP1x+dtWS8McKP1P4hcKUwv8WGHEUli5fPmp0pExhN4V7K+yvcLDC4QrHAI6JTFA4VeHhCmcqPFbh8QoXLL/w/OWR0xWepbBK4QUKL1F4hcLrVEbLi4a9ioadi4RlkDv5kk+JfI8NUhrVeWsv1Zid/gXSGVqpWNi9aEg+aL7r6PhiuhQNexQNexYNexcNuxYN9y4a7lM07FYU7A/tPRPz0Xy6kVbRPbSGnsS6/SV6jWrpXfqQPsEKOkX/5BIuw8q5Lw/mkTyOZ/NxvAAr3qV8gRkfnmjDCTY8WmfRnshZxSt5Pb/F252OTqVzuLPQudhZ6TzqvOS843zq/HeoLNQ/NCo0LXR0aIGW4dBkG0634RwbnmrD8214lQ1X2fBxG9bY8EMbfmvCcFjnZQ53MjiGL7PhJTa82IYX2dD2LfyADd+y4btaX2lkaGRqZE5kSeSSyI2RByLPmtTIBhtutuFrptXIB+a55HQbopWSlfQNOYAdeRH9mfvQ56B0OWg8gQ/jU0Ddy5HnRv1aa4eSVXncSq1DrpDWWYV8F+RxVZojWjKnkZuNUrN9NQxHrjGN3HBNb1/SJcd1RImO2dKR7dQ+siPHbdfU0si7AQdawmdLbkCO6oDboGntIvf43B3IeYev1OVIv8bnLteUksg5nluMXIt9JWYjdY7nZmt8JDLRugOQ4wBf7r5IG2RdX40NR8rUlSK1NJszvIPC4W/V7dC4UPgTuI+Q8pEvVy3i34Gr1Rgn/EoYXAKfzfE4Yp8GlGcOr1afTb1RvUm7Dv4KE5Ls8Zocyzy+dmNORY3nh88JxM0GHgvgjg/EjkMfpqubFogHFcIjrasMpISpJNzNc2X+tNCn1C70jc99GUj9LZWGPgi4dwLpa6l9qDrHbQjkuJ6ioTsauZu9PByC9IQWk5k3+mnMHHhQJ3S6jeNQFea2TrBzp8MKPguSGsXMsbfuNoVoEz/PN2r4gs6Dm7iab9DwJZQP8002zXx/OGZz/NKXY6Xv/mb33vkt3wJYwz8DfInvBnyeHyHH+QK2RU/aTLfR/VgtDNLV1hBY2fvBph8Ge7wXvYqeyS7581gHmF1xx6mhXvwwr+F1fCs/wA/xHXwn/4rv4tV8N9/D9/J9fD8/xo/yWl7Ft/MjfBs/iFKvoK8P0IPqHWcrDePPeTt/wl9xiv/OO/gLTvLX/CV/zH/hv3GC4/wp/5U/azF+oLqzFnQuNaE7Es4q+B2YJ36TjWOMrwOJca7yxd2H59+Z0Iv7BTy0La/21fcknqGn+XpfPsg+xo+dh3z5IC0Mfe+stHGOI99qepfr+H3+gP+A5zuohD/i9/iPXM8f8p8Qs4o6IOZdxNlcXho7t5JZnfUHHUZh3TVNV7LzsBZaTOfQ+XQRXUZX8VVo+0Say1dreBJfo+ECvlbDk/k6DRfyTzU8hf8sIdq7Arw1l68EPBE9c1DyZ4ALeAXgyfxzwIWghoMSDeTwe8BV7LNN/Dg/wU/yf2DefYqf5md4A29E+jdUxv/AGD/Lz4GzXwCvbuLN/CK/xFtkfNXWWkCyX/8zupvepL/Qt5jzB/J4PphP4rP438QO4wo+W3MO1C+LyxsaY6VtkmfZGyfSL41vgiO0tUVTxdJhUNrg9RQ/o7lv5XVaupdyTQ9wVy/ddWjMW7InMFo573lw4KsiUVh7HooV7umoSXYAxebqD+cAs4Gg31i4MNa38g28KoxEKWybR5DzMVoL6X6ansGoyfuCzopFV5SLUTFvqeQ9VIm+h+qv7yGG6Buo/fTd0wiqobex4pW3TlP0rdM0fesklkQI+aVV6fcdfL/2ey9rSUNy+Aa+kX8JW+kmvhly/KBiMJ9n8izYUEfhqQQYnMIn8Dw+kefjuQOe5yPG5vClMHjI0REarmMR1h0jQx0zYo5i0pHvRfIA8Psb1IcfAobvUTteQ3vzidAp71N7vodKuY4O4bdpPG+mctgjvZ2nqYtzI/yLWDN9S/P4OdoPcjMVtiPBfjkSFuAwfpm68dPU13mD9nFkt+wJxPVSH+FDqbeUoW00mjuBmnU0lYeBc3vSUO5PPfgx2gc49OQPIHUPA4+/oty1GKXtNAj3g+kjOgR+NK+iIfwqjUU4hk+jHqFONIIzKPMZ8P2A9nUGInwZ/jE6xOmO59Eo9z7BrqLuvAPxDyD+YfiR8BNoX34X4USEcaQJDV6kMmcQcPuaOjh/oE6Qnm5KE6FDP+SpBE790JcraCB3z/wTI3oC34m0t2kcNEwfhP3AV324A8X0y2lv0jh6mcbT1swboLvcT3CeQ17EQwv20XJvoUx/hEL/gSj3azoAnNDBqaK90M+9dDwwBjwFfPMV2puC++Opr/ZbPPotfXb7pPgLTvm84Lgu6IET8Mr8Ff5L+P8Cbvsrbrle8PL79cDhh1TJP0Hai8DhQYzjGIzPDjzXoz/nUVfnHhroDAUd76dHMSYLnYZMBjiWQtf15GWg5d1U7twHvH4DHjwdvCW8tDjzMHRdBXTcIOc8w2NuX6V/oQj6+DHaOBO8Ngc4TAH9pkBfPAPN/hQNV94UmmHc6I1MFX+CFc92SIDw0rM0ACucqcg7DGWGhnuAllK38Iwbvmx4ht7P/A/CEHyp0tX1oK/raStk5H46Dn4W/LnwC+HXwF8Kvxb+JPiL4OfyXXQ7/Ab4u/H8Ovw58Cfb8An4E62XOh6xdcyyz6+At0t0jLsDN+En4bVaw18eL7h0Ap2VL3K8Jx85XmXF7w+imCs7oEMaYTl8R9x/prL0tZEl9Q9auRqZ4/00M75U5cz1Im+5HvLn935aBzxk0+9VRl05dX0O7yufCf/neldmG/towL8JPlR5zqRA807whPu4J9+rjXyrfxX6NkUTVM4/yjRo+KZP7l3/uobtVP6Nr1Q9kOtFL/h9rixaLzrD71UOXP1hvfN7lP8HxUJb4TdjnJ8wPvRzhE/Cv4X0V4x37rXhg9D9j8C/A93+Bmj5J9rbWYe6YBPy7dDdm6AnbwJ9H8EYfEoDuJy6wmrpybch/AB13IrwNejMq6Ej3kb4a5Stpd7hdsDpXdBJ/Id0MP8e4X/CPwVaPEVTIK89eB54722McT/QoTfaOAj3vRE/Efc90cZEvf/u8h0BK+UZzENvwYr5C2QRod9LnOe/BY/8DXK8DTSJg67Qj7BY91G91x9j+BMa4iyGdeHqqxVIX4GwAjp5IHiwP+SmP2g2DbhMQ1hJXXgIePEbmu88jnlyJtH/ArfDUkIKZW5kc3RyZWFtCmVuZG9iagoyMSAwIG9iago1MDkwMQplbmRvYmoKMjAgMCBvYmoKMTI2NDY0CmVuZG9iagoxNiAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDkxID4+CnN0cmVhbQp4nGNgoBAIA7EomCUBJqUZZOByikCszKDCoAqk1YFYE4i1sZpihsS2IdEFIQyhQDKcIYIhkiEKyIphiAWS8QwJDIkMSQzJQHYqQxpDOkMGQyZDFpCXAwDAkAl0CmVuZHN0cmVhbQplbmRvYmoKMTkgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAzNzQgPj4Kc3RyZWFtCnicXZNPb4MwDMXvfIoct0NFoTRZJYQ0dRcO+6OxnaodKDEV0ggo0EO//ZI8Q6Uhtdb7+bl2UhMfy5fSdLOIP+zQVDSLtjPa0jRcbUPiTJfOREkqdNfMrMJ309djFLvi6jbN1JemHaI8F/GnS06zvYmHZz2c6TESQsTvVpPtzEU8fB8roOo6jr/Uk5nFNioKoal1P/daj291TyIOxZtSu3w33zau7O74uo0k0qATjNQMmqaxbsjW5kJRvnVPIfLWPUVERv/LJwpl53b1p96PcEL8CTgFThmnjPfAe8Z7xk/AB8arRJYgibMEvEPnHbdYZLJQmDLIjE0ZYwksGUvGaOzDCTHgbBewDydE4AMwT53xuBITSJ7rLpHFQFJxliXuRcolCS8ayJq9q0S2CVLxzbOUGgF3JtvFE0oU/hXFzReJ4ymMojCKkosVleit+KzKn9Xtx7IIflX8Xq972FytdSsYlj/snt+6ztD6fozD6Kv85w+zktm4CmVuZHN0cmVhbQplbmRvYmoKMTQgMCBvYmoKPDwgL0Jhc2VGb250IC9HdWFyZGlhblNhbnNDb25kLVJlZ3VsYXIKL0NJRFN5c3RlbUluZm8gPDwgL09yZGVyaW5nIChJZGVudGl0eSkgL1JlZ2lzdHJ5IChBZG9iZSkgL1N1cHBsZW1lbnQgMCA+PgovQ0lEVG9HSURNYXAgMTYgMCBSIC9Gb250RGVzY3JpcHRvciAxMyAwIFIgL1N1YnR5cGUgL0NJREZvbnRUeXBlMgovVHlwZSAvRm9udCAvVyAxOCAwIFIgPj4KZW5kb2JqCjE1IDAgb2JqCjw8IC9CYXNlRm9udCAvR3VhcmRpYW5TYW5zQ29uZC1SZWd1bGFyIC9EZXNjZW5kYW50Rm9udHMgWyAxNCAwIFIgXQovRW5jb2RpbmcgL0lkZW50aXR5LUggL1N1YnR5cGUgL1R5cGUwIC9Ub1VuaWNvZGUgMTkgMCBSIC9UeXBlIC9Gb250ID4+CmVuZG9iagoxMyAwIG9iago8PCAvQXNjZW50IDgwOSAvQ2FwSGVpZ2h0IDAgL0Rlc2NlbnQgLTE5MSAvRmxhZ3MgMzIKL0ZvbnRCQm94IFsgLTE0NiAtMjE2IDEwMTggMTA0NCBdIC9Gb250RmlsZTIgMTcgMCBSCi9Gb250TmFtZSAvR3VhcmRpYW5TYW5zQ29uZC1SZWd1bGFyIC9JdGFsaWNBbmdsZSAwIC9NYXhXaWR0aCA3MTUgL1N0ZW1WIDAKL1R5cGUgL0ZvbnREZXNjcmlwdG9yIC9YSGVpZ2h0IDAgPj4KZW5kb2JqCjE4IDAgb2JqClsgMzIgWyAxNTggXSAzNCBbIDMxMiBdIDM3IFsgNzE1IF0gNDAgWyAyODAgMjgwIF0gNDYgWyAyMTYgXSA0OApbIDUzNSAyODggNDI1IF0gNTIgWyA0NzQgXSA1NCBbIDQ4OSBdIDU2IFsgNDk0IF0gNjcgWyA0NzQgXSA3MyBbIDIxNyBdIDk3ClsgNDA5IDQ2NiBdIDEwMCBbIDQ2NiA0MjcgMjY3IDQyNyBdIDEwNSBbIDE5MyAxOTIgXSAxMDgKWyAxOTMgNjk1IDQ2MSA0NTYgNDY2IF0gMTE0IFsgMjg4IDM0NSAyOTAgNDU1IDQxOSA2MjIgXSAxMjEgWyA0MTQgXSBdCmVuZG9iagozIDAgb2JqCjw8IC9GMSAxNSAwIFIgPj4KZW5kb2JqCjQgMCBvYmoKPDwgL0ExIDw8IC9DQSAwIC9UeXBlIC9FeHRHU3RhdGUgL2NhIDEgPj4KL0EyIDw8IC9DQSAxIC9UeXBlIC9FeHRHU3RhdGUgL2NhIDEgPj4gPj4KZW5kb2JqCjUgMCBvYmoKPDwgPj4KZW5kb2JqCjYgMCBvYmoKPDwgPj4KZW5kb2JqCjcgMCBvYmoKPDwgPj4KZW5kb2JqCjIgMCBvYmoKPDwgL0NvdW50IDEgL0tpZHMgWyAxMSAwIFIgXSAvVHlwZSAvUGFnZXMgPj4KZW5kb2JqCjIyIDAgb2JqCjw8IC9DcmVhdGlvbkRhdGUgKEQ6MjAyMTEwMjMxNzEwMDUrMDInMDAnKQovQ3JlYXRvciAoTWF0cGxvdGxpYiB2My40LjMsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcpCi9Qcm9kdWNlciAoTWF0cGxvdGxpYiBwZGYgYmFja2VuZCB2My40LjMpID4+CmVuZG9iagp4cmVmCjAgMjMKMDAwMDAwMDAwMCA2NTUzNSBmIAowMDAwMDAwMDE2IDAwMDAwIG4gCjAwMDAwNTQxNjkgMDAwMDAgbiAKMDAwMDA1Mzk3NSAwMDAwMCBuIAowMDAwMDU0MDA3IDAwMDAwIG4gCjAwMDAwNTQxMDYgMDAwMDAgbiAKMDAwMDA1NDEyNyAwMDAwMCBuIAowMDAwMDU0MTQ4IDAwMDAwIG4gCjAwMDAwMDAwNjUgMDAwMDAgbiAKMDAwMDAwMDQwMSAwMDAwMCBuIAowMDAwMDAxNDIzIDAwMDAwIG4gCjAwMDAwMDAyMDggMDAwMDAgbiAKMDAwMDAwMTQwMyAwMDAwMCBuIAowMDAwMDUzNDY1IDAwMDAwIG4gCjAwMDAwNTMwOTEgMDAwMDAgbiAKMDAwMDA1MzMxMSAwMDAwMCBuIAowMDAwMDUyNDgxIDAwMDAwIG4gCjAwMDAwMDE0NDMgMDAwMDAgbiAKMDAwMDA1MzY5NSAwMDAwMCBuIAowMDAwMDUyNjQ0IDAwMDAwIG4gCjAwMDAwNTI0NTggMDAwMDAgbiAKMDAwMDA1MjQzNiAwMDAwMCBuIAowMDAwMDU0MjI5IDAwMDAwIG4gCnRyYWlsZXIKPDwgL0luZm8gMjIgMCBSIC9Sb290IDEgMCBSIC9TaXplIDIzID4+CnN0YXJ0eHJlZgo1NDM4NgolJUVPRgo=\n", - "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-23T17:10:05.056089\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", - "text/plain": [ - "
" - ] - }, - "metadata": { - "needs_background": "light" - }, - "output_type": "display_data" - } - ], - "source": [ - "preds_df = pd.DataFrame(preds[0])\n", - "plt.bar(labels, 100 * preds_df[\"score\"], color='C0')\n", - "plt.title(f'\"{custom_tweet}\"')\n", - "plt.ylabel(\"Class probability (%)\")\n", - "plt.show()" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## Conclusion" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "Congratulations, you now know how to train a transformer model to classify the emotions in tweets! We have seen two complementary approaches based on features and fine-tuning, and investigated their strengths and weaknesses. \n", - "\n", - "However, this is just the first step in building a real-world application with transformer models, and we have a lot more ground to cover. Here's a list of challenges you're likely to experience in your NLP journey:\n", - "\n", - "My boss wants my model in production yesterday!::\n", - "In most applications, your model doesn't just sit somewhere gathering dust - you want to make sure it's serving predictions! When a model is pushed to the Hub, an inference endpoint is automatically created that can be called with HTTP requests. We recommend checking out the [documentation](https://api-inference.huggingface.co/docs/python/html/index.html) of the Inference API if you want to learn more. \n", - "\n", - "My users want faster predictions!::\n", - "We've already seen one approach to this problem: using DistilBERT. In <> we'll dive into knowledge distillation (the process by which DistilBERT was created), along with other tricks to speed up your transformer models.\n", - "\n", - "\n", - "Can your model also do X?::\n", - "As we've alluded to in this chapter, transformers are extremely versatile. In the rest of the book we will be exploring a range of tasks, like question answering and named entity recognition, all using the same basic architecture.\n", - "\n", - "None of my texts are in English!::\n", - "It turns out that transformers also come in a multilingual variety, and we'll use them in <> to tackle several languages at once.\n", - "\n", - "I don't have any labels!::\n", - "If there is very little labeled data available, fine-tuning may not be an option. In <>, we'll explore some techniques to deal with this situation.\n", - "\n", - "Now that we've seen what's involved in training and sharing a transformer, in the next chapter we'll explore implementing our very own transformer model from scratch." - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [] - } - ], - "metadata": { - "kernelspec": { - "display_name": "book", - "language": "python", - "name": "python3" - } - }, - "nbformat": 4, - "nbformat_minor": 4 -} +{ + "cells": [ + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Uncomment and run this cell if you're on Colab or Kaggle\n", + "# !git clone https://github.com/nlp-with-transformers/notebooks.git\n", + "# %cd notebooks\n", + "# from install import *\n", + "# install_requirements(is_chapter2=True)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# hide\n", + "from utils import *\n", + "setup_chapter()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Text Classification" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Text classification is one of the most common tasks in NLP; it can be used for a broad range of applications, such as tagging customer feedback into categories or routing support tickets according to their language. Chances are that your email program's spam filter is using text classification to protect your inbox from a deluge of unwanted junk!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Another common type of text classification is sentiment analysis, which (as we saw in <>) aims to identify the polarity of a given text. For example, a company like Tesla might analyze Twitter posts like the one in <> to determine whether people like its new car roofs or not." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\"Tesla" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now imagine that you are a data scientist who needs to build a system that can automatically identify emotional states such as \"anger\" or \"joy\" that people express about your company's product on Twitter. In this chapter, we'll tackle this task using a variant of BERT called DistilBERT.footnote:[V. Sanh et al., [\"DistilBERT, a Distilled Version of BERT: Smaller, Faster, Cheaper and Lighter\"](https://arxiv.org/abs/1910.01108), (2019).] The main advantage of this model is that it achieves comparable performance to BERT, while being significantly smaller and more efficient. This enables us to train a classifier in a few minutes, and if you want to train a larger BERT model you can simply change the checkpoint of the pretrained model. A _checkpoint_ corresponds to the set of weights that are loaded into a given transformer architecture.\n", + "\n", + "This will also be our first encounter with three of the core libraries from the Hugging Face ecosystem: image:images/logo.png[hf,13,13] Datasets, image:images/logo.png[hf,13,13] Tokenizers, and image:images/logo.png[hf,13,13] Transformers. As shown in <>, these libraries will allow us to quickly go from raw text to a fine-tuned model that can be used for inference on new tweets. So, in the spirit of Optimus Prime, let's dive in, \"transform, and roll out!\"footnote:[Optimus Prime is the leader of a race of robots in the popular Transformers franchise for children (and for those who are young at heart!).]" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\"Hugging" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## The Dataset" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To build our emotion detector we'll use a great dataset from an article that explored how emotions are represented in English Twitter messages.footnote:[E. Saravia et al., \"CARER: Contextualized Affect Representations for Emotion Recognition,\" _Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing_ (Oct–Nov 2018): 3687–3697, http://dx.doi.org/10.18653/v1/D18-1404.] Unlike most sentiment analysis datasets that involve just \"positive\" and \"negative\" polarities, this dataset contains six basic emotions: anger, disgust, fear, joy, sadness, and surprise. Given a tweet, our task will be to train a model that can classify it into one of these emotions." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### A First Look at Hugging Face Datasets" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We will use image:images/logo.png[hf,13,13] Datasets to download the data from the [Hugging Face Hub](https://huggingface.co/datasets). We can use the `list_datasets()` function to see what datasets are available on the Hub:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "There are 1753 datasets currently available on the Hub\n", + "The first 10 are: ['acronym_identification', 'ade_corpus_v2', 'adversarial_qa',\n", + "'aeslc', 'afrikaans_ner_corpus', 'ag_news', 'ai2_arc', 'air_dialogue',\n", + "'ajgt_twitter_ar', 'allegro_reviews']\n" + ] + } + ], + "source": [ + "from datasets import list_datasets\n", + "\n", + "all_datasets = list_datasets()\n", + "print(f\"There are {len(all_datasets)} datasets currently available on the Hub\")\n", + "print(f\"The first 10 are: {all_datasets[:10]}\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We see that each dataset is given a name, so let's load the `emotion` dataset with the `load_dataset()` function:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "1454a965e1f5435fbf369734d6608857", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Downloading: 0%| | 0.00/1.66k [00:00>." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "```asciidoc\n", + "[[dataset-loading]]\n", + ".How to load datasets in various formats\n", + "[options=\"header\"]\n", + "|======\n", + "| Data format | Loading script | Example\n", + "| CSV | `csv` | `load_dataset(\"csv\", data_files=\"my_file.csv\")` \n", + "| Text | `text` | `load_dataset(\"text\", data_files=\"my_file.txt\")` \n", + "| JSON | `json` | `load_dataset(\"json\", data_files=\"my_file.jsonl\")`\n", + "|======\n", + "```" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As you can see for each data format, we just need to pass the relevant loading script to the `load_dataset()` function, along with a `data_files` argument that specifies the path or URL to one or more files. For example, the source files for the `emotion` dataset are actually hosted on Dropbox, so an alternative way to load the dataset is to first download one of the splits:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "--2023-02-14 19:24:47-- https://huggingface.co/datasets/transformersbook/emotion-train-split/raw/main/train.txt\n", + "Resolving huggingface.co (huggingface.co)... 54.235.118.239, 3.231.67.228, 2600:1f18:147f:e850:e203:c458:10cd:fc3c, ...\n", + "Connecting to huggingface.co (huggingface.co)|54.235.118.239|:443... connected.\n", + "HTTP request sent, awaiting response... 200 OK\n", + "Length: 1658616 (1,6M) [text/plain]\n", + "Saving to: ‘train.txt’\n", + "\n", + "train.txt 100%[===================>] 1,58M 1,06MB/s in 1,5s \n", + "\n", + "2023-02-14 19:24:49 (1,06 MB/s) - ‘train.txt’ saved [1658616/1658616]\n", + "\n" + ] + } + ], + "source": [ + "#hide_output\n", + "# The original URL used in the book is no longer available, so we use a different one\n", + "dataset_url = \"https://huggingface.co/datasets/transformersbook/emotion-train-split/raw/main/train.txt\"\n", + "!wget {dataset_url}" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "If you’re wondering why there’s a `!` character in the preceding shell command, that’s because we’re running the commands in a Jupyter notebook. Simply remove the prefix if you want to download and unzip the dataset within a terminal. Now, if we peek at the first row of the _train.txt_ file:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "i didnt feel humiliated;sadness\n" + ] + } + ], + "source": [ + "!head -n 1 train.txt" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "we can see that here are no column headers and each tweet and emotion are separated by a semicolon. Nevertheless, this is quite similar to a CSV file, so we can load the dataset locally by using the `csv` script and pointing the `data_files` argument to the _train.txt_ file:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Using custom data configuration default-dd8fa13a78374240\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Downloading and preparing dataset csv/default to /Users/lewtun/.cache/huggingface/datasets/csv/default-dd8fa13a78374240/0.0.0/bf68a4c4aefa545d0712b2fcbb1b327f905bbe2f6425fbc5e8c25234acb9e14a...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "946eefc885d64620a0a05968dcd939f3", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + " 0%| | 0/1 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
textlabel
0i didnt feel humiliated0
1i can go from feeling so hopeless to so damned...0
2im grabbing a minute to post i feel greedy wrong3
3i am ever feeling nostalgic about the fireplac...2
4i am feeling grouchy3
\n", + "" + ], + "text/plain": [ + " text label\n", + "0 i didnt feel humiliated 0\n", + "1 i can go from feeling so hopeless to so damned... 0\n", + "2 im grabbing a minute to post i feel greedy wrong 3\n", + "3 i am ever feeling nostalgic about the fireplac... 2\n", + "4 i am feeling grouchy 3" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "import pandas as pd\n", + "\n", + "emotions.set_format(type=\"pandas\")\n", + "df = emotions[\"train\"][:]\n", + "df.head()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As you can see, the column headers have been preserved and the first few rows match our previous views of the data. However, the labels are represented as integers, so let's use the `int2str()` method of the `label` feature to create a new column in our `DataFrame` with the corresponding label names:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
textlabellabel_name
0i didnt feel humiliated0sadness
1i can go from feeling so hopeless to so damned...0sadness
2im grabbing a minute to post i feel greedy wrong3anger
3i am ever feeling nostalgic about the fireplac...2love
4i am feeling grouchy3anger
\n", + "
" + ], + "text/plain": [ + " text label label_name\n", + "0 i didnt feel humiliated 0 sadness\n", + "1 i can go from feeling so hopeless to so damned... 0 sadness\n", + "2 im grabbing a minute to post i feel greedy wrong 3 anger\n", + "3 i am ever feeling nostalgic about the fireplac... 2 love\n", + "4 i am feeling grouchy 3 anger" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "def label_int2str(row):\n", + " return emotions[\"train\"].features[\"label\"].int2str(row)\n", + "\n", + "df[\"label_name\"] = df[\"label\"].apply(label_int2str)\n", + "df.head()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Before diving into building a classifier, let's take a closer look at the dataset. As Andrej Karpathy notes in his famous blog post [\"A Recipe for Training Neural Networks\"](https://karpathy.github.io/2019/04/25/recipe), becoming \"one with the data\" is an essential step for training great models!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Looking at the Class Distribution" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Whenever you are working on text classification problems, it is a good idea to examine the distribution of examples across the classes. A dataset with a skewed class distribution might require a different treatment in terms of the training loss and evaluation metrics than a balanced one. \n", + "\n", + "With Pandas and Matplotlib, we can quickly visualize the class distribution as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-29T16:17:44.966027\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "import matplotlib.pyplot as plt\n", + "\n", + "df[\"label_name\"].value_counts(ascending=True).plot.barh()\n", + "plt.title(\"Frequency of Classes\")\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this case, we can see that the dataset is heavily imbalanced; the `joy` and `sadness` classes appear frequently, whereas `love` and `surprise` are about 5–10 times rarer. There are several ways to deal with imbalanced data, including:\n", + "\n", + "* Randomly oversample the minority class.\n", + "* Randomly undersample the majority class.\n", + "* Gather more labeled data from the underrepresented classes." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To keep things simple in this chapter, we'll work with the raw, unbalanced class frequencies. If you want to learn more about these sampling techniques, we recommend checking out the [Imbalanced-learn library](https://imbalanced-learn.org/stable/). Just make sure that you don't apply sampling methods _before_ creating your train/test splits, or you'll get plenty of leakage between them!\n", + "\n", + "Now that we've looked at the classes, let's take a look at the tweets themselves." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### How Long Are Our Tweets?" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Transformer models have a maximum input sequence length that is referred to as the _maximum context size_. For applications using DistilBERT, the maximum context size is 512 tokens, which amounts to a few paragraphs of text. As we'll see in the next section, a token is an atomic piece of text; for now, we'll treat a token as a single word. We can get a rough estimate of tweet lengths per emotion by looking at the distribution of words per tweet:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-29T16:18:00.302888\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "df[\"Words Per Tweet\"] = df[\"text\"].str.split().apply(len)\n", + "df.boxplot(\"Words Per Tweet\", by=\"label_name\", grid=False, showfliers=False,\n", + " color=\"black\")\n", + "plt.suptitle(\"\")\n", + "plt.xlabel(\"\")\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "From the plot we see that for each emotion, most tweets are around 15 words long and the longest tweets are well below DistilBERT's maximum context size. Texts that are longer than a model's context size need to be truncated, which can lead to a loss in performance if the truncated text contains crucial information; in this case, it looks like that won't be an issue. " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Let's now figure out how we can convert these raw texts into a format suitable for image:images/logo.png[hf,13,13] Transformers! While we're at it, let's also reset the output format of our dataset since we don't need the `DataFrame` format anymore: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "emotions.reset_format()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## From Text to Tokens" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Transformer models like DistilBERT cannot receive raw strings as input; instead, they assume the text has been _tokenized_ and _encoded_ as numerical vectors. Tokenization is the step of breaking down a string into the atomic units used in the model. There are several tokenization strategies one can adopt, and the optimal splitting of words into subunits is usually learned from the corpus. Before looking at the tokenizer used for DistilBERT, let's consider two extreme cases: _character_ and _word_ tokenization." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Character Tokenization" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The simplest tokenization scheme is to feed each character individually to the model. In Python, `str` objects are really arrays under the hood, which allows us to quickly implement character-level tokenization with just one line of code:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "['T', 'o', 'k', 'e', 'n', 'i', 'z', 'i', 'n', 'g', ' ', 't', 'e', 'x', 't', ' ',\n", + "'i', 's', ' ', 'a', ' ', 'c', 'o', 'r', 'e', ' ', 't', 'a', 's', 'k', ' ', 'o',\n", + "'f', ' ', 'N', 'L', 'P', '.']\n" + ] + } + ], + "source": [ + "text = \"Tokenizing text is a core task of NLP.\"\n", + "tokenized_text = list(text)\n", + "print(tokenized_text)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This is a good start, but we're not done yet. Our model expects each character to be converted to an integer, a process sometimes called _numericalization_. One simple way to do this is by encoding each unique token (which are characters in this case) with a unique integer:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "{' ': 0, '.': 1, 'L': 2, 'N': 3, 'P': 4, 'T': 5, 'a': 6, 'c': 7, 'e': 8, 'f': 9,\n", + "'g': 10, 'i': 11, 'k': 12, 'n': 13, 'o': 14, 'r': 15, 's': 16, 't': 17, 'x': 18,\n", + "'z': 19}\n" + ] + } + ], + "source": [ + "token2idx = {ch: idx for idx, ch in enumerate(sorted(set(tokenized_text)))}\n", + "print(token2idx)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This gives us a mapping from each character in our vocabulary to a unique integer. We can now use `token2idx` to transform the tokenized text to a list of integers:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "[5, 14, 12, 8, 13, 11, 19, 11, 13, 10, 0, 17, 8, 18, 17, 0, 11, 16, 0, 6, 0, 7,\n", + "14, 15, 8, 0, 17, 6, 16, 12, 0, 14, 9, 0, 3, 2, 4, 1]\n" + ] + } + ], + "source": [ + "input_ids = [token2idx[token] for token in tokenized_text]\n", + "print(input_ids)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Each token has now been mapped to a unique numerical identifier (hence the name `input_ids`). The last step is to convert `input_ids` to a 2D tensor of one-hot vectors. One-hot vectors are frequently used in machine learning to encode categorical data, which can be either ordinal or nominal. For example, suppose we wanted to encode the names of characters in the _Transformers_ TV series. One way to do this would be to map each name to a unique ID, as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
NameLabel ID
0Bumblebee0
1Optimus Prime1
2Megatron2
\n", + "
" + ], + "text/plain": [ + " Name Label ID\n", + "0 Bumblebee 0\n", + "1 Optimus Prime 1\n", + "2 Megatron 2" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "categorical_df = pd.DataFrame(\n", + " {\"Name\": [\"Bumblebee\", \"Optimus Prime\", \"Megatron\"], \"Label ID\": [0,1,2]})\n", + "categorical_df" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The problem with this approach is that it creates a fictitious ordering between the names, and neural networks are _really_ good at learning these kinds of relationships. So instead, we can create a new column for each category and assign a 1 where the category is true, and a 0 otherwise. In Pandas, this can be implemented with the `get_dummies()` function as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
BumblebeeMegatronOptimus Prime
0100
1001
2010
\n", + "
" + ], + "text/plain": [ + " Bumblebee Megatron Optimus Prime\n", + "0 1 0 0\n", + "1 0 0 1\n", + "2 0 1 0" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "pd.get_dummies(categorical_df[\"Name\"])" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The rows of this `DataFrame` are the one-hot vectors, which have a single \"hot\" entry with a 1 and 0s everywhere else. Now, looking at our `input_ids`, we have a similar problem: the elements create an ordinal scale. This means that adding or subtracting two IDs is a meaningless operation, since the result is a new ID that represents another random token.\n", + "\n", + "On the other hand, the result of adding two one-hot encodings can easily be interpreted: the two entries that are \"hot\" indicate that the corresponding tokens co-occur. We can create the one-hot encodings in PyTorch by converting `input_ids` to a tensor and applying the `one_hot()` function as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "torch.Size([38, 20])" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "import torch\n", + "import torch.nn.functional as F\n", + "\n", + "input_ids = torch.tensor(input_ids)\n", + "one_hot_encodings = F.one_hot(input_ids, num_classes=len(token2idx))\n", + "one_hot_encodings.shape" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "For each of the 38 input tokens we now have a one-hot vector with 20 dimensions, since our vocabulary consists of 20 unique characters." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> Warning: It's important to always set `num_classes` in the `one_hot()` function because otherwise the one-hot vectors may end up being shorter than the length of the vocabulary (and need to be padded with zeros manually). In TensorFlow, the equivalent function is `tf.one_hot()`, where the `depth` argument plays the role of `num_classes`." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "By examining the first vector, we can verify that a 1 appears in the location indicated by `input_ids[0]`:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Token: T\n", + "Tensor index: 5\n", + "One-hot: tensor([0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0])\n" + ] + } + ], + "source": [ + "print(f\"Token: {tokenized_text[0]}\")\n", + "print(f\"Tensor index: {input_ids[0]}\")\n", + "print(f\"One-hot: {one_hot_encodings[0]}\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "From our simple example we can see that character-level tokenization ignores any structure in the text and treats the whole string as a stream of characters. Although this helps deal with misspellings and rare words, the main drawback is that linguistic structures such as words need to be _learned_ from the data. This requires significant compute, memory, and data. For this reason, character tokenization is rarely used in practice. Instead, some structure of the text is preserved during the tokenization step. _Word tokenization_ is a straightforward approach to achieve this, so let's take a look at how it works." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Word Tokenization" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Instead of splitting the text into characters, we can split it into words and map each word to an integer. Using words from the outset enables the model to skip the step of learning words from characters, and thereby reduces the complexity of the training process." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "One simple class of word tokenizers uses whitespace to tokenize the text. We can do this by applying Python's `split()` function directly on the raw text (just like we did to measure the tweet lengths):" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "['Tokenizing', 'text', 'is', 'a', 'core', 'task', 'of', 'NLP.']\n" + ] + } + ], + "source": [ + "tokenized_text = text.split()\n", + "print(tokenized_text)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "From here we can take the same steps we took for the character tokenizer to map each word to an ID. However, we can already see one potential problem with this tokenization scheme: punctuation is not accounted for, so `NLP.` is treated as a single token. Given that words can include declinations, conjugations, or misspellings, the size of the vocabulary can easily grow into the millions! \n", + "\n", + "\n", + "> note: Some word tokenizers have extra rules for punctuation. One can also apply stemming or lemmatization, which normalizes words to their stem (e.g., \"great\", \"greater\", and \"greatest\" all become \"great\"), at the expense of losing some information in the text. " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Having a large vocabulary is a problem because it requires neural networks to have an enormous number of parameters. To illustrate this, suppose we have 1 million unique words and want to compress the 1-million-dimensional input vectors to 1-thousand-dimensional vectors in the first layer of our neural network. This is a standard step in most NLP architectures, and the resulting weight matrix of this first layer would contain 1 million $\\times$ 1 thousand = 1 billion weights. This is already comparable to the largest GPT-2 model,footnote:[GPT-2 is the successor of GPT, and it captivated the public's attention with its impressive ability to generate realistic text. We'll explore GPT-2 in detail in <>.] which has around 1.5 billion parameters in total!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Naturally, we want to avoid being so wasteful with our model parameters since models are expensive to train, and larger models are more difficult to maintain. A common approach is to limit the vocabulary and discard rare words by considering, say, the 100,000 most common words in the corpus. Words that are not part of the vocabulary are classified as \"unknown\" and mapped to a shared `UNK` token. This means that we lose some potentially important information in the process of word tokenization, since the model has no information about words associated with `UNK`.\n", + "\n", + "Wouldn't it be nice if there was a compromise between character and word tokenization that preserved all the input information _and_ some of the input structure? There is: _subword tokenization_." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Subword Tokenization" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The basic idea behind subword tokenization is to combine the best aspects of character and word tokenization. On the one hand, we want to split rare words into smaller units to allow the model to deal with complex words and misspellings. On the other hand, we want to keep frequent words as unique entities so that we can keep the length of our inputs to a manageable size. The main distinguishing feature of subword tokenization (as well as word tokenization) is that it is _learned_ from the pretraining corpus using a mix of statistical rules and algorithms.\n", + "\n", + "There are several subword tokenization algorithms that are commonly used in NLP, but let's start with WordPiece,footnote:[M. Schuster and K. Nakajima, \"Japanese and Korean Voice Search,\" _2012 IEEE International Conference on Acoustics, Speech and Signal Processing_ (2012): 5149–5152, https://doi.org/10.1109/ICASSP.2012.6289079.] which is used by the BERT and DistilBERT tokenizers. The easiest way to understand how WordPiece works is to see it in action. image:images/logo.png[hf,13,13] Transformers provides a convenient `AutoTokenizer` class that allows you to quickly load the tokenizer associated with a pretrained model—we just call its `from_pretrained()` method, providing the ID of a model on the Hub or a local file path. Let's start by loading the tokenizer for DistilBERT:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# hide_output\n", + "from transformers import AutoTokenizer\n", + "\n", + "model_ckpt = \"distilbert-base-uncased\"\n", + "tokenizer = AutoTokenizer.from_pretrained(model_ckpt)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The `AutoTokenizer` class belongs to a larger set of [\"auto\" classes](https://huggingface.co/docs/transformers/model_doc/auto) whose job is to automatically retrieve the model's configuration, pretrained weights, or vocabulary from the name of the checkpoint. This allows you to quickly switch between models, but if you wish to load the specific class manually you can do so as well. For example, we could have loaded the DistilBERT tokenizer as follows:\n", + "\n", + "```python\n", + "from transformers import DistilBertTokenizer\n", + "\n", + "distilbert_tokenizer = DistilBertTokenizer.from_pretrained(model_ckpt)\n", + "```" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> note: When you run the `AutoTokenizer.from_pretrained()` method for the first time you will see a progress bar that shows which parameters of the pretrained tokenizer are loaded from the Hugging Face Hub. When you run the code a second time, it will load the tokenizer from the cache, usually located at _~/.cache/huggingface/_." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Let's examine how this tokenizer works by feeding it our simple \"Tokenizing text is a core task of NLP.\" example text:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "{'input_ids': [101, 19204, 6026, 3793, 2003, 1037, 4563, 4708, 1997, 17953,\n", + "2361, 1012, 102], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}\n" + ] + } + ], + "source": [ + "encoded_text = tokenizer(text)\n", + "print(encoded_text)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Just like we saw with character tokenization, we can see that the words have been mapped to unique integers in the `input_ids` field. We'll discuss the role of the `attention_mask` field in the next section. Now that we have the `input_ids`, we can convert them back into tokens by using the tokenizer's `convert_ids_to_tokens()` method:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "['[CLS]', 'token', '##izing', 'text', 'is', 'a', 'core', 'task', 'of', 'nl',\n", + "'##p', '.', '[SEP]']\n" + ] + } + ], + "source": [ + "tokens = tokenizer.convert_ids_to_tokens(encoded_text.input_ids)\n", + "print(tokens)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can observe three things here. First, some special `[CLS]` and `[SEP]` tokens have been added to the start and end of the sequence. These tokens differ from model to model, but their main role is to indicate the start and end of a sequence. Second, the tokens have each been lowercased, which is a feature of this particular checkpoint. Finally, we can see that \"tokenizing\" and \"NLP\" have been split into two tokens, which makes sense since they are not common words. The `##` prefix in `##izing` and `##p` means that the preceding string is not whitespace; any token with this prefix should be merged with the previous token when you convert the tokens back to a string. The `AutoTokenizer` class has a `convert_tokens_to_string()` method for doing just that, so let's apply it to our tokens:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "[CLS] tokenizing text is a core task of nlp. [SEP]\n" + ] + } + ], + "source": [ + "print(tokenizer.convert_tokens_to_string(tokens))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The `AutoTokenizer` class also has several attributes that provide information about the tokenizer. For example, we can inspect the vocabulary size:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "30522" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "tokenizer.vocab_size" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "and the corresponding model's maximum context size:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "512" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "tokenizer.model_max_length" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Another interesting attribute to know about is the names of the fields that the model expects in its forward pass:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "['input_ids', 'attention_mask']" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "tokenizer.model_input_names" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now that we have a basic understanding of the tokenization process for a single string, let's see how we can tokenize the whole dataset!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> warning: When using pretrained models, it is _really_ important to make sure that you use the same tokenizer that the model was trained with. From the model's perspective, switching the tokenizer is like shuffling the vocabulary. If everyone around you started swapping random words like \"house\" for \"cat,\" you'd have a hard time understanding what was going on too!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Tokenizing the Whole Dataset" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To tokenize the whole corpus, we'll use the `map()` method of our `DatasetDict` object. We'll encounter this method many times throughout this book, as it provides a convenient way to apply a processing function to each element in a dataset. As we'll soon see, the `map()` method can also be used to create new rows and columns.\n", + "\n", + "To get started, the first thing we need is a processing function to tokenize our examples with:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "def tokenize(batch):\n", + " return tokenizer(batch[\"text\"], padding=True, truncation=True)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This function applies the tokenizer to a batch of examples; `padding=True` will pad the examples with zeros to the size of the longest one in a batch, and `truncation=True` will truncate the examples to the model's maximum context size. To see `tokenize()` in action, let's pass a batch of two examples from the training set:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "{'input_ids': [[101, 1045, 2134, 2102, 2514, 26608, 102, 0, 0, 0, 0, 0, 0, 0, 0,\n", + "0, 0, 0, 0, 0, 0, 0, 0], [101, 1045, 2064, 2175, 2013, 3110, 2061, 20625, 2000,\n", + "2061, 9636, 17772, 2074, 2013, 2108, 2105, 2619, 2040, 14977, 1998, 2003, 8300,\n", + "102]], 'attention_mask': [[1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n", + "0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n", + "1, 1]]}\n" + ] + } + ], + "source": [ + "print(tokenize(emotions[\"train\"][:2]))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Here we can see the result of padding: the first element of `input_ids` is shorter than the second, so zeros have been added to that element to make them the same length. These zeros have a corresponding `[PAD]` token in the vocabulary, and the set of special tokens also includes the `[CLS]` and `[SEP]` tokens that we encountered earlier:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
Special Token[PAD][UNK][CLS][SEP][MASK]
Special Token ID0100101102103
" + ], + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "#hide_input\n", + "tokens2ids = list(zip(tokenizer.all_special_tokens, tokenizer.all_special_ids))\n", + "data = sorted(tokens2ids, key=lambda x : x[-1])\n", + "df = pd.DataFrame(data, columns=[\"Special Token\", \"Special Token ID\"])\n", + "df.T" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Also note that in addition to returning the encoded tweets as `input_ids`, the tokenizer returns a list of `attention_mask` arrays. This is because we do not want the model to get confused by the additional padding tokens: the attention mask allows the model to ignore the padded parts of the input. <> provides a visual explanation of how the input IDs and attention masks are padded.\n", + "\n", + "\"attention-mask\" " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Once we've defined a processing function, we can apply it across all the splits in the corpus in a single line of code:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# hide_output\n", + "emotions_encoded = emotions.map(tokenize, batched=True, batch_size=None)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "By default, the `map()` method operates individually on every example in the corpus, so setting `batched=True` will encode the tweets in batches. Because we've set `batch_size=None`, our `tokenize()` function will be applied on the full dataset as a single batch. This ensures that the input tensors and attention masks have the same shape globally, and we can see that this operation has added new `input_ids` and `attention_mask` columns to the dataset:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "['attention_mask', 'input_ids', 'label', 'text']\n" + ] + } + ], + "source": [ + "print(emotions_encoded[\"train\"].column_names)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> Note: In later chapters, we'll see how _data collators_ can be used to dynamically pad the tensors in each batch. Padding globally will come in handy in the next section, where we extract a feature matrix from the whole corpus." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Training a Text Classifier" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As discussed in <>, models like DistilBERT are pretrained to predict masked words in a sequence of text. However, we can't use these language models directly for text classification; we need to modify them slightly. To understand what modifications are necessary, let's take a look at the architecture of an encoder-based model like DistilBERT, which is depicted in <>. " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\"encoder-classifier\"" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "First, the text is tokenized and represented as one-hot vectors called _token encodings_. The size of the tokenizer vocabulary determines the dimension of the token encodings, and it usually consists of 20k–200k unique tokens. Next, these token encodings are converted to _token embeddings_, which are vectors living in a lower-dimensional space. The token embeddings are then passed through the encoder block layers to yield a _hidden state_ for each input token. For the pretraining objective of language modeling,⁠footnote:[In the case of DistilBERT, it's guessing the masked tokens.] each hidden state is fed to a layer that predicts the masked input tokens. For the classification task, we replace the language modeling layer with a classification layer." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> note: In practice, PyTorch skips the step of creating one-hot vectors for token encodings because multiplying a matrix with a one-hot vector is the same as selecting a column from the matrix. This can be done directly by getting the column with the token ID from the matrix. We'll see this in <> when we use the `nn.Embedding` class." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We have two options to train such a model on our Twitter dataset:\n", + "\n", + "- _Feature extraction_:: We use the hidden states as features and just train a classifier on them, without modifying the pretrained model.\n", + "- _Fine-tuning_:: We train the whole model end-to-end, which also updates the parameters of the pretrained model. \n", + "\n", + "In the following sections we explore both options for DistilBERT and examine their trade-offs. " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Transformers as Feature Extractors" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "Using a transformer as a feature extractor is fairly simple. As shown in <>, we freeze the body's weights during training and use the hidden states as features for the classifier. The advantage of this approach is that we can quickly train a small or shallow model. Such a model could be a neural classification layer or a method that does not rely on gradients, such as a random forest. This method is especially convenient if GPUs are unavailable, since the hidden states only need to be precomputed once." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\"encoder-features\"" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Using pretrained models" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "We will use another convenient auto class from image:images/logo.png[hf,13,13] Transformers called `AutoModel`. Similar to the `AutoTokenizer` class, `AutoModel` has a `from_pretrained()` method to load the weights of a pretrained model. Let's use this method to load the DistilBERT checkpoint:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# hide_output\n", + "from transformers import AutoModel\n", + "\n", + "model_ckpt = \"distilbert-base-uncased\"\n", + "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n", + "model = AutoModel.from_pretrained(model_ckpt).to(device)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Here we've used PyTorch to check whether a GPU is available or not, and then chained the PyTorch `nn.Module.to()` method to the model loader. This ensures that the model will run on the GPU if we have one. If not, the model will run on the CPU, which can be considerably slower." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The `AutoModel` class converts the token encodings to embeddings, and then feeds them through the encoder stack to return the hidden states. Let's take a look at how we can extract these states from our corpus." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Sidebar: Interoperability Between Frameworks" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Although the code in this book is mostly written in PyTorch, image:images/logo.png[hf,13,13] Transformers provides tight interoperability with TensorFlow and JAX. This means that you only need to change a few lines of code to load a pretrained model in your favorite deep learning framework! For example, we can load DistilBERT in TensorFlow by using the `TFAutoModel` class as follows: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2021-10-23 17:03:51.654626: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory\n", + "2021-10-23 17:03:51.654664: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1835] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n", + "Skipping registering GPU devices...\n", + "2021-10-23 17:03:51.655491: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA\n", + "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n", + "2021-10-23 17:03:51.680031: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n" + ] + } + ], + "source": [ + "#hide_output\n", + "from transformers import TFAutoModel\n", + "\n", + "tf_model = TFAutoModel.from_pretrained(model_ckpt)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This interoperability is especially useful when a model is only released in one framework, but you'd like to use it in another. For example, the [XLM-RoBERTa model](https://huggingface.co/xlm-roberta-base) that we'll encounter in <> only has PyTorch weights, so if you try to load it in TensorFlow as we did before:\n", + "\n", + "```python\n", + "tf_xlmr = TFAutoModel.from_pretrained(\"xlm-roberta-base\")\n", + "```\n", + "\n", + "you'll get an error. In these cases, you can specify a `from_pt=True` argument to the `TfAutoModel.from_pretrained()` function, and the library will automatically download and convert the PyTorch weights for you:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "tf_xlmr = TFAutoModel.from_pretrained(\"xlm-roberta-base\", from_pt=True)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As you can see, it is very simple to switch between frameworks in image:images/logo.png[hf,13,13] Transformers! In most cases, you can just add a \"TF\" prefix to the classes and you'll get the equivalent TensorFlow 2.0 classes. When we use the `\"pt\"` string (e.g., in the following section), which is short for PyTorch, just replace it with \"`tf\"`, which is short for TensorFlow." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### End sidebar" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Extracting the last hidden states" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To warm up, let's retrieve the last hidden states for a single string. The first thing we need to do is encode the string and convert the tokens to PyTorch tensors. This can be done by providing the `return_tensors=\"pt\"` argument to the tokenizer as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Input tensor shape: torch.Size([1, 6])\n" + ] + } + ], + "source": [ + "text = \"this is a test\"\n", + "inputs = tokenizer(text, return_tensors=\"pt\")\n", + "print(f\"Input tensor shape: {inputs['input_ids'].size()}\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As we can see, the resulting tensor has the shape `[batch_size, n_tokens]`. Now that we have the encodings as a tensor, the final step is to place them on the same device as the model and pass the inputs as follows: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "BaseModelOutput(last_hidden_state=tensor([[[-0.1565, -0.1862, 0.0528, ...,\n", + "-0.1188, 0.0662, 0.5470],\n", + " [-0.3575, -0.6484, -0.0618, ..., -0.3040, 0.3508, 0.5221],\n", + " [-0.2772, -0.4459, 0.1818, ..., -0.0948, -0.0076, 0.9958],\n", + " [-0.2841, -0.3917, 0.3753, ..., -0.2151, -0.1173, 1.0526],\n", + " [ 0.2661, -0.5094, -0.3180, ..., -0.4203, 0.0144, -0.2149],\n", + " [ 0.9441, 0.0112, -0.4714, ..., 0.1439, -0.7288, -0.1619]]],\n", + " device='cuda:0'), hidden_states=None, attentions=None)\n" + ] + } + ], + "source": [ + "inputs = {k:v.to(device) for k,v in inputs.items()}\n", + "with torch.no_grad():\n", + " outputs = model(**inputs)\n", + "print(outputs)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Here we've used the `torch.no_grad()` context manager to disable the automatic calculation of the gradient. This is useful for inference since it reduces the memory footprint of the computations. Depending on the model configuration, the output can contain several objects, such as the hidden states, losses, or attentions, arranged in a class similar to a `namedtuple` in Python. In our example, the model output is an instance of `BaseModelOutput`, and we can simply access its attributes by name. The current model returns only one attribute, which is the last hidden state, so let's examine its shape:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "torch.Size([1, 6, 768])" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "outputs.last_hidden_state.size()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Looking at the hidden state tensor, we see that it has the shape `[batch_size, n_tokens, hidden_dim]`. In other words, a 768-dimensional vector is returned for each of the 6 input tokens. For classification tasks, it is common practice to just use the hidden state associated with the `[CLS]` token as the input feature. Since this token appears at the start of each sequence, we can extract it by simply indexing into `outputs.last_hidden_state` as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "torch.Size([1, 768])" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "outputs.last_hidden_state[:,0].size()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now that we know how to get the last hidden state for a single string, let's do the same thing for the whole dataset by creating a new `hidden_state` column that stores all these vectors. As we did with the tokenizer, we'll use the `map()` method of `DatasetDict` to extract all the hidden states in one go. The first thing we need to do is wrap the previous steps in a processing function:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "def extract_hidden_states(batch):\n", + " # Place model inputs on the GPU\n", + " inputs = {k:v.to(device) for k,v in batch.items() \n", + " if k in tokenizer.model_input_names}\n", + " # Extract last hidden states\n", + " with torch.no_grad():\n", + " last_hidden_state = model(**inputs).last_hidden_state\n", + " # Return vector for [CLS] token\n", + " return {\"hidden_state\": last_hidden_state[:,0].cpu().numpy()}" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The only difference between this function and our previous logic is the final step where we place the final hidden state back on the CPU as a NumPy array. The `map()` method requires the processing function to return Python or NumPy objects when we're using batched inputs.\n", + "\n", + "Since our model expects tensors as inputs, the next thing to do is convert the `input_ids` and `attention_mask` columns to the `\"torch\"` format, as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "emotions_encoded.set_format(\"torch\", \n", + " columns=[\"input_ids\", \"attention_mask\", \"label\"])" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can then go ahead and extract the hidden states across all splits in one go:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "590842bb15bf448cb35e324e87fdadd9", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + " 0%| | 0/16 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
XYlabel
04.3580756.1408160
1-3.1345675.3294460
25.1522302.7326433
3-2.5190183.0672502
4-3.3645203.3566133
\n", + "" + ], + "text/plain": [ + " X Y label\n", + "0 4.358075 6.140816 0\n", + "1 -3.134567 5.329446 0\n", + "2 5.152230 2.732643 3\n", + "3 -2.519018 3.067250 2\n", + "4 -3.364520 3.356613 3" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from umap import UMAP\n", + "from sklearn.preprocessing import MinMaxScaler\n", + "\n", + "# Scale features to [0,1] range\n", + "X_scaled = MinMaxScaler().fit_transform(X_train)\n", + "# Initialize and fit UMAP\n", + "mapper = UMAP(n_components=2, metric=\"cosine\").fit(X_scaled)\n", + "# Create a DataFrame of 2D embeddings\n", + "df_emb = pd.DataFrame(mapper.embedding_, columns=[\"X\", \"Y\"])\n", + "df_emb[\"label\"] = y_train\n", + "df_emb.head()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The result is an array with the same number of training samples, but with only 2 features instead of the 768 we started with! Let's investigate the compressed data a little bit further and plot the density of points for each category separately:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-23T17:07:41.645879\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "fig, axes = plt.subplots(2, 3, figsize=(7,5))\n", + "axes = axes.flatten()\n", + "cmaps = [\"Greys\", \"Blues\", \"Oranges\", \"Reds\", \"Purples\", \"Greens\"]\n", + "labels = emotions[\"train\"].features[\"label\"].names\n", + "\n", + "for i, (label, cmap) in enumerate(zip(labels, cmaps)):\n", + " df_emb_sub = df_emb.query(f\"label == {i}\")\n", + " axes[i].hexbin(df_emb_sub[\"X\"], df_emb_sub[\"Y\"], cmap=cmap,\n", + " gridsize=20, linewidths=(0,))\n", + " axes[i].set_title(label)\n", + " axes[i].set_xticks([]), axes[i].set_yticks([])\n", + "\n", + "plt.tight_layout()\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + ">note: These are only projections onto a lower-dimensional space. Just because some categories overlap does not mean that they are not separable in the original space. Conversely, if they are separable in the projected space they will be separable in the original space.\n", + "\n", + "From this plot we can see some clear patterns: the negative feelings such as `sadness`, `anger`, and `fear` all occupy similar regions with slightly varying distributions. On the other hand, `joy` and `love` are well separated from the negative emotions and also share a similar space. Finally, `surprise` is scattered all over the place. Although we may have hoped for some separation, this is in no way guaranteed since the model was not trained to know the difference between these emotions. It only learned them implicitly by guessing the masked words in texts.\n", + "\n", + "Now that we've gained some insight into the features of our dataset, let's finally train a model on it!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Training a simple classifier\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We've seen that the hidden states are somewhat different between the emotions, although for several of them there is no obvious boundary. Let's use these hidden states to train a logistic regression model with Scikit-Learn. Training such a simple model is fast and does not require a GPU:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/home/lewis/miniconda3/envs/book/lib/python3.7/site-packages/sklearn/linear_model/_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n", + "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n", + "\n", + "Increase the number of iterations (max_iter) or scale the data as shown in:\n", + " https://scikit-learn.org/stable/modules/preprocessing.html\n", + "Please also refer to the documentation for alternative solver options:\n", + " https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n", + " extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n" + ] + }, + { + "data": { + "text/plain": [ + "LogisticRegression()" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "#hide_output\n", + "# We increase `max_iter` to guarantee convergence \n", + "from sklearn.linear_model import LogisticRegression\n", + "\n", + "lr_clf = LogisticRegression(max_iter=3000)\n", + "lr_clf.fit(X_train, y_train)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "0.6085" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "lr_clf.score(X_valid, y_valid)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Looking at the accuracy, it might appear that our model is just a bit better than random—but since we are dealing with an unbalanced multiclass dataset, it's actually significantly better. We can examine whether our model is any good by comparing it against a simple baseline. In Scikit-Learn there is a `DummyClassifier` that can be used to build a classifier with simple heuristics such as always choosing the majority class or always drawing a random class. In this case the best-performing heuristic is to always choose the most frequent class, which yields an accuracy of about 35%:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "0.352" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from sklearn.dummy import DummyClassifier\n", + "\n", + "dummy_clf = DummyClassifier(strategy=\"most_frequent\")\n", + "dummy_clf.fit(X_train, y_train)\n", + "dummy_clf.score(X_valid, y_valid)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "So, our simple classifier with DistilBERT embeddings is significantly better than our baseline. We can further investigate the performance of the model by looking at the confusion matrix of the classifier, which tells us the relationship between the true and predicted labels:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-12-07T11:46:25.597214\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix\n", + "\n", + "def plot_confusion_matrix(y_preds, y_true, labels):\n", + " cm = confusion_matrix(y_true, y_preds, normalize=\"true\")\n", + " fig, ax = plt.subplots(figsize=(6, 6))\n", + " disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=labels)\n", + " disp.plot(cmap=\"Blues\", values_format=\".2f\", ax=ax, colorbar=False)\n", + " plt.title(\"Normalized confusion matrix\")\n", + " plt.show()\n", + " \n", + "y_preds = lr_clf.predict(X_valid)\n", + "plot_confusion_matrix(y_preds, y_valid, labels)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can see that `anger` and `fear` are most often confused with `sadness`, which agrees with the observation we made when visualizing the embeddings. Also, `love` and `surprise` are frequently mistaken for `joy`." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In the next section we will explore the fine-tuning approach, which leads to superior classification performance. It is, however, important to note that doing this requires more computational resources, such as GPUs, that might not be available in your organization. In cases like these, a feature-based approach can be a good compromise between doing traditional machine learning and deep learning." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Fine-Tuning Transformers" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "Let's now explore what it takes to fine-tune a transformer end-to-end. With the fine-tuning approach we do not use the hidden states as fixed features, but instead train them as shown in <>. This requires the classification head to be differentiable, which is why this method usually uses a neural network for classification.\n", + "\n", + "\"encoder-tuning\"" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Training the hidden states that serve as inputs to the classification model will help us avoid the problem of working with data that may not be well suited for the classification task. Instead, the initial hidden states adapt during training to decrease the model loss and thus increase its performance.\n", + "\n", + "We'll be using the `Trainer` API from image:images/logo.png[hf,13,13] Transformers to simplify the training loop. Let's look at the ingredients we need to set one up!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Loading a pretrained model" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The first thing we need is a pretrained DistilBERT model like the one we used in the feature-based approach. The only slight modification is that we use the `AutoModelForSequenceClassification` model instead of `AutoModel`. The difference is that the `AutoModelForSequenceClassification` model has a classification head on top of the pretrained model outputs, which can be easily trained with the base model. We just need to specify how many labels the model has to predict (six in our case), since this dictates the number of outputs the classification head has:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# hide_output\n", + "from transformers import AutoModelForSequenceClassification\n", + "\n", + "num_labels = 6\n", + "model = (AutoModelForSequenceClassification\n", + " .from_pretrained(model_ckpt, num_labels=num_labels)\n", + " .to(device))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "You will see a warning that some parts of the model are randomly initialized. This is normal since the classification head has not yet been trained. The next step is to define the metrics that we'll use to evaluate our model's performance during fine-tuning." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Defining the performance metrics" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "To monitor metrics during training, we need to define a `compute_metrics()` function for the `Trainer`. This function receives an `EvalPrediction` object (which is a named tuple with `predictions` and `label_ids` attributes) and needs to return a dictionary that maps each metric's name to its value. For our application, we'll compute the $F_1$-score and the accuracy of the model as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "from sklearn.metrics import accuracy_score, f1_score\n", + "\n", + "def compute_metrics(pred):\n", + " labels = pred.label_ids\n", + " preds = pred.predictions.argmax(-1)\n", + " f1 = f1_score(labels, preds, average=\"weighted\")\n", + " acc = accuracy_score(labels, preds)\n", + " return {\"accuracy\": acc, \"f1\": f1}" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "With the dataset and metrics ready, we just have two final things to take care of before we define the `Trainer` class:\n", + "\n", + "1. Log in to our account on the Hugging Face Hub. This will allow us to push our fine-tuned model to our account on the Hub and share it with the community.\n", + "2. Define all the hyperparameters for the training run.\n", + "\n", + "We'll tackle these steps in the next section." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Training the model" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "If you're running this code in a Jupyter notebook, you can log in to the Hub with the following helper function:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "from huggingface_hub import notebook_login\n", + "\n", + "notebook_login()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This will display a widget in which you can enter your username and password, or an access token with write privileges. You can find details on how to create access tokens in the [Hub documentation](https://huggingface.co/docs/hub/security#user-access-tokens). If you're working in the terminal, you can log in by running the following command:\n", + "\n", + "```bash\n", + "$ huggingface-cli login\n", + "```" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To define the training parameters, we use the `TrainingArguments` class. This class stores a lot of information and gives you fine-grained control over the training and evaluation. The most important argument to specify is `output_dir`, which is where all the artifacts from training are stored. Here is an example of `TrainingArguments` in all its glory:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "from transformers import Trainer, TrainingArguments\n", + "\n", + "batch_size = 64\n", + "logging_steps = len(emotions_encoded[\"train\"]) // batch_size\n", + "model_name = f\"{model_ckpt}-finetuned-emotion\"\n", + "training_args = TrainingArguments(output_dir=model_name,\n", + " num_train_epochs=2,\n", + " learning_rate=2e-5,\n", + " per_device_train_batch_size=batch_size,\n", + " per_device_eval_batch_size=batch_size,\n", + " weight_decay=0.01,\n", + " evaluation_strategy=\"epoch\",\n", + " disable_tqdm=False,\n", + " logging_steps=logging_steps,\n", + " push_to_hub=True, \n", + " log_level=\"error\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Here we also set the batch size, learning rate, and number of epochs, and specify to load the best model at the end of the training run. With this final ingredient, we can instantiate and fine-tune our model with the `Trainer`: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "\n", + "
\n", + " \n", + " \n", + " [500/500 01:48, Epoch 2/2]\n", + "
\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
EpochTraining LossValidation LossAccuracyF1
10.8409000.3274450.8965000.892285
20.2550000.2204720.9225000.922550

" + ], + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "from transformers import Trainer\n", + "\n", + "trainer = Trainer(model=model, args=training_args, \n", + " compute_metrics=compute_metrics,\n", + " train_dataset=emotions_encoded[\"train\"],\n", + " eval_dataset=emotions_encoded[\"validation\"],\n", + " tokenizer=tokenizer)\n", + "trainer.train();" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Looking at the logs, we can see that our model has an $F_1$-score on the validation set of around 92% - this is a significant improvement over the feature-based approach!\n", + "\n", + "We can take a more detailed look at the training metrics by calculating the confusion matrix. To visualize the confusion matrix, we first need to get the predictions on the validation set. The `predict()` method of the `Trainer` class returns several useful objects we can use for evaluation:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "\n", + "

\n", + " \n", + " \n", + " [32/32 00:01]\n", + "
\n", + " " + ], + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# hide_output\n", + "preds_output = trainer.predict(emotions_encoded[\"validation\"])" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The output of the `predict()` method is a `PredictionOutput` object that contains arrays of `predictions` and `label_ids`, along with the metrics we passed to the trainer. For example, the metrics on the validation set can be accessed as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "{'test_loss': 0.22047173976898193,\n", + " 'test_accuracy': 0.9225,\n", + " 'test_f1': 0.9225500751072866,\n", + " 'test_runtime': 1.6357,\n", + " 'test_samples_per_second': 1222.725,\n", + " 'test_steps_per_second': 19.564}" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "preds_output.metrics" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "It also contains the raw predictions for each class. We can decode the predictions greedily using `np.argmax()`. This yields the predicted labels and has the same format as the labels returned by the Scikit-Learn models in the feature-based approach:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "y_preds = np.argmax(preds_output.predictions, axis=1)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "With the predictions, we can plot the confusion matrix again:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-12-07T11:59:17.791697\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "plot_confusion_matrix(y_preds, y_valid, labels)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "This is much closer to the ideal diagonal confusion matrix. The `love` category is still often confused with `joy`, which seems natural. `surprise` is also frequently mistaken for `joy`, or confused with `fear`. Overall the performance of the model seems quite good, but before we call it a day, let's dive a little deeper into the types of errors our model is likely to make." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Sidebar: Fine-Tuning with Keras" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "If you are using TensorFlow, it's also possible to fine-tune your models using the Keras API. The main difference from the PyTorch API is that there is no `Trainer` class, since Keras models already provide a built-in `fit()` method. To see how this works, let's first load DistilBERT as a TensorFlow model:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2021-10-29 15:33:36.938811: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory\n", + "2021-10-29 15:33:36.938844: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1835] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n", + "Skipping registering GPU devices...\n", + "2021-10-29 15:33:36.939933: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA\n", + "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n", + "2021-10-29 15:33:36.962642: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n" + ] + } + ], + "source": [ + "#hide_output\n", + "from transformers import TFAutoModelForSequenceClassification\n", + "\n", + "tf_model = (TFAutoModelForSequenceClassification\n", + " .from_pretrained(model_ckpt, num_labels=num_labels))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Next, we'll convert our datasets into the `tf.data.Dataset` format. Since we have already padded our tokenized inputs, we can do this easily by applying the `to_tf_dataset()` method to `emotions_encoded`:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# The column names to convert to TensorFlow tensors\n", + "tokenizer_columns = tokenizer.model_input_names\n", + "\n", + "tf_train_dataset = emotions_encoded[\"train\"].to_tf_dataset(\n", + " columns=tokenizer_columns, label_cols=[\"label\"], shuffle=True,\n", + " batch_size=batch_size)\n", + "tf_eval_dataset = emotions_encoded[\"validation\"].to_tf_dataset(\n", + " columns=tokenizer_columns, label_cols=[\"label\"], shuffle=False,\n", + " batch_size=batch_size)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Here we've also shuffled the training set, and defined the batch size for it and the validation set. The last thing to do is compile and train the model:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2021-10-29 15:36:00.548707: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Epoch 1/2\n", + "250/250 [==============================] - 478s 2s/step - loss: 0.5379 - sparse_categorical_accuracy: 0.8138 - val_loss: 0.1452 - val_sparse_categorical_accuracy: 0.9430\n", + "Epoch 2/2\n", + "250/250 [==============================] - 471s 2s/step - loss: 0.1424 - sparse_categorical_accuracy: 0.9415 - val_loss: 0.1512 - val_sparse_categorical_accuracy: 0.9335\n" + ] + }, + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "#hide_output\n", + "import tensorflow as tf\n", + "\n", + "tf_model.compile(\n", + " optimizer=tf.keras.optimizers.Adam(learning_rate=5e-5),\n", + " loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n", + " metrics=tf.metrics.SparseCategoricalAccuracy())\n", + "\n", + "tf_model.fit(tf_train_dataset, validation_data=tf_eval_dataset, epochs=2)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### End sidebar" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Error analysis" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Before moving on, we should investigate our model's predictions a little bit further. A simple yet powerful technique is to sort the validation samples by the model loss. When we pass the label during the forward pass, the loss is automatically calculated and returned. Here's a function that returns the loss along with the predicted label:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "from torch.nn.functional import cross_entropy\n", + "\n", + "def forward_pass_with_label(batch):\n", + " # Place all input tensors on the same device as the model\n", + " inputs = {k:v.to(device) for k,v in batch.items() \n", + " if k in tokenizer.model_input_names}\n", + "\n", + " with torch.no_grad():\n", + " output = model(**inputs)\n", + " pred_label = torch.argmax(output.logits, axis=-1)\n", + " loss = cross_entropy(output.logits, batch[\"label\"].to(device), \n", + " reduction=\"none\")\n", + "\n", + " # Place outputs on CPU for compatibility with other dataset columns \n", + " return {\"loss\": loss.cpu().numpy(), \n", + " \"predicted_label\": pred_label.cpu().numpy()}" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Using the `map()` method once more, we can apply this function to get the losses for all the samples:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "6004443ac1344ee18d40c8a90c1178f4", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + " 0%| | 0/125 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
textlabelpredicted_labelloss
1801i feel that he was being overshadowed by the s...lovesadness5.704531
1963i called myself pro life and voted for perry w...joysadness5.484461
1870i guess i feel betrayed because i admired him ...joysadness5.434768
882i feel badly about reneging on my commitment t...lovesadness5.257482
1950i as representative of everything thats wrong ...surprisesadness4.827708
1509i guess this is a memoir so it feels like that...joyfear4.713047
1274i am going to several holiday parties and i ca...joysadness4.704955
318i felt ashamed of these feelings and was scare...fearsadness4.656096
1500i guess we would naturally feel a sense of lon...angersadness4.593202
1111im lazy my characters fall into categories of ...joyfear4.311287
\n", + "" + ], + "text/plain": [ + " text label \\\n", + "1801 i feel that he was being overshadowed by the s... love \n", + "1963 i called myself pro life and voted for perry w... joy \n", + "1870 i guess i feel betrayed because i admired him ... joy \n", + "882 i feel badly about reneging on my commitment t... love \n", + "1950 i as representative of everything thats wrong ... surprise \n", + "1509 i guess this is a memoir so it feels like that... joy \n", + "1274 i am going to several holiday parties and i ca... joy \n", + "318 i felt ashamed of these feelings and was scare... fear \n", + "1500 i guess we would naturally feel a sense of lon... anger \n", + "1111 im lazy my characters fall into categories of ... joy \n", + "\n", + " predicted_label loss \n", + "1801 sadness 5.704531 \n", + "1963 sadness 5.484461 \n", + "1870 sadness 5.434768 \n", + "882 sadness 5.257482 \n", + "1950 sadness 4.827708 \n", + "1509 fear 4.713047 \n", + "1274 sadness 4.704955 \n", + "318 sadness 4.656096 \n", + "1500 sadness 4.593202 \n", + "1111 fear 4.311287 " + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "#hide_output\n", + "df_test.sort_values(\"loss\", ascending=False).head(10)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can clearly see that the model predicted some of the labels incorrectly. On the other hand, it seems that there are quite a few examples with no clear class, which might be either mislabeled or require a new class altogether. In particular, `joy` seems to be mislabeled several times. With this information we can refine the dataset, which often can lead to as big a performance gain (or more) as having more data or larger models!" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "When looking at the samples with the lowest losses, we observe that the model seems to be most confident when predicting the `sadness` class. Deep learning models are exceptionally good at finding and exploiting shortcuts to get to a prediction. For this reason, it is also worth investing time into looking at the examples that the model is most confident about, so that we can be confident that the model does not improperly exploit certain features of the text. So, let's also look at the predictions with the smallest loss:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
textlabelpredicted_labelloss
21i feel try to tell me im ungrateful tell me im...sadnesssadness0.017331
244im kinda relieve but at the same time i feel d...sadnesssadness0.017392
133i and feel quite ungrateful for it but i m loo...sadnesssadness0.017400
392i remember feeling disheartened one day when w...sadnesssadness0.017461
1310i feel like an ungrateful assholesadnesssadness0.017485
189i leave the meeting feeling more than a little...sadnesssadness0.017670
1120i am feeling a little disheartenedsadnesssadness0.017685
783i feel like i deserve to be broke with how fri...sadnesssadness0.017888
1368i started this blog with pure intentions i mus...sadnesssadness0.017899
1466i feel so ungrateful to be wishing this pregna...sadnesssadness0.017913
\n", + "
" + ], + "text/plain": [ + " text label \\\n", + "21 i feel try to tell me im ungrateful tell me im... sadness \n", + "244 im kinda relieve but at the same time i feel d... sadness \n", + "133 i and feel quite ungrateful for it but i m loo... sadness \n", + "392 i remember feeling disheartened one day when w... sadness \n", + "1310 i feel like an ungrateful asshole sadness \n", + "189 i leave the meeting feeling more than a little... sadness \n", + "1120 i am feeling a little disheartened sadness \n", + "783 i feel like i deserve to be broke with how fri... sadness \n", + "1368 i started this blog with pure intentions i mus... sadness \n", + "1466 i feel so ungrateful to be wishing this pregna... sadness \n", + "\n", + " predicted_label loss \n", + "21 sadness 0.017331 \n", + "244 sadness 0.017392 \n", + "133 sadness 0.017400 \n", + "392 sadness 0.017461 \n", + "1310 sadness 0.017485 \n", + "189 sadness 0.017670 \n", + "1120 sadness 0.017685 \n", + "783 sadness 0.017888 \n", + "1368 sadness 0.017899 \n", + "1466 sadness 0.017913 " + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "#hide_output\n", + "df_test.sort_values(\"loss\", ascending=True).head(10)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We now know that the `joy` is sometimes mislabeled and that the model is most confident about predicting the label `sadness`. With this information we can make targeted improvements to our dataset, and also keep an eye on the class the model seems to be very confident about. \n", + "\n", + "The last step before serving the trained model is to save it for later usage. image:images/logo.png[hf,13,13] Transformers allows us to do this in a few steps, which we'll show you in the next section." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "#### Saving and sharing the model" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "The NLP community benefits greatly from sharing pretrained and fine-tuned models, and everybody can share their models with others via the Hugging Face Hub. Any community-generated model can be downloaded from the Hub just like we downloaded the DistilBERT model. With the `Trainer` API, saving and sharing a model is simple:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "'https://huggingface.co/lewtun/distilbert-base-uncased-finetuned-emotion/commit/352c4147e4754f73a0b41f7b175f4a907270c9c9'" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "#hide_output\n", + "trainer.push_to_hub(commit_message=\"Training completed!\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can also use the fine-tuned model to make predictions on new tweets. Since we've pushed our model to the Hub, we can now use it with the `pipeline()` function, just like we did in <>. First, let's load the pipeline:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "#hide_output\n", + "from transformers import pipeline\n", + "\n", + "# Change `transformersbook` to your Hub username\n", + "model_id = \"transformersbook/distilbert-base-uncased-finetuned-emotion\"\n", + "classifier = pipeline(\"text-classification\", model=model_id)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Then let's test the pipeline with a sample tweet:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "custom_tweet = \"I saw a movie today and it was really good.\"\n", + "preds = classifier(custom_tweet, return_all_scores=True)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Finally, we can plot the probability for each class in a bar plot. Clearly, the model estimates that the most likely class is `joy`, which appears to be reasonable given the tweet:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "JVBERi0xLjQKJazcIKu6CjEgMCBvYmoKPDwgL1BhZ2VzIDIgMCBSIC9UeXBlIC9DYXRhbG9nID4+CmVuZG9iago4IDAgb2JqCjw8IC9FeHRHU3RhdGUgNCAwIFIgL0ZvbnQgMyAwIFIgL1BhdHRlcm4gNSAwIFIKL1Byb2NTZXQgWyAvUERGIC9UZXh0IC9JbWFnZUIgL0ltYWdlQyAvSW1hZ2VJIF0gL1NoYWRpbmcgNiAwIFIKL1hPYmplY3QgNyAwIFIgPj4KZW5kb2JqCjExIDAgb2JqCjw8IC9Bbm5vdHMgMTAgMCBSIC9Db250ZW50cyA5IDAgUgovR3JvdXAgPDwgL0NTIC9EZXZpY2VSR0IgL1MgL1RyYW5zcGFyZW5jeSAvVHlwZSAvR3JvdXAgPj4KL01lZGlhQm94IFsgMCAwIDM4Ny45MzQzNzUgMjY3LjA1ODc1IF0gL1BhcmVudCAyIDAgUiAvUmVzb3VyY2VzIDggMCBSCi9UeXBlIC9QYWdlID4+CmVuZG9iago5IDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggMTIgMCBSID4+CnN0cmVhbQp4nL1WTW8bRwzd8/yKQYAC8aFjkvN9jNPEaG9pBfSSixLLig3ZbuUkRv5938wmXq4iCT0JxkI7z8N5JJd8HLa35vwV2/WjJXuL58myvbTnv62+3nxc/Xl5YT8+GgJ+Z3zJrvrgc8Ryo5eSsqNY8LbBXr36ZMy9wemwuMTBa2NCfLaKLuKnHUwuz7DNDAveETdQ2T5jYLg2/9qfjvU+uGKFswvBblf2b3sPJwgrzz5LDNU2hv5GYrdw0Rz5rzGJHUeJMU+OM3nH3udQJsc1VpwIiU+AlbVCu/Pv7CncZ/YuSaWQlP8xusTwJSv/FebFpUSZQ8OVvcZPGUJKjilGr2pHqDgqKUuaQtBYcXBJsm8RKHMFnzAA4eJiwccPKoBELiYvrIpfY8nl5Jlqg5W5gk8ZQObW2pG96l4UBvmaKajuVVhyJcHn5r+yntATuu/FO3x+wnJyP+GNE1ev3FdYdpmiLy3R2lzBzwGcv5JRR9eQOmgpnH/qwgfEFHEMz6KiVpA46szmAmr5ZC4W9vwtWxa7uDZwxsdApVap0VYXamriurgyL4fHYTlcDffDCm+PZ3Zxa98sTCc2jOwhhBRYNbzCjlFyE6kQsC/EnMEZY/U/OG+Hh+HbDhdmAdWky0JBR5lycYxGLBRKrDvRbcD0dVjNucTDHfS2SqSCjnEJvlqEslEmRgZgw/KdaYkcrpHF7Q5XxXmhtj6byBR2lK34XiMo+CB+J7JrcC132XwIrs2IWlR5KuwYm0dsqXAJOYL0pyr5MmyHf/Dc4H0nn3sGchDI43we7+ccMyhoa68KhA4RpIqhDQ/DjGNCDyQS/c8tlhQctqXKRaSOVHKYjBkMMUR8J82m4AN0kCYExYTrDUc0QeuBkS0cYcPYLDVI8nO2CT7IlsaNjC6PQhiwJY106QhdrQ7ySJzndBN8kC5WSi2+GlyuocaQEeZIWA4TCkgy1DTLvD4m+FDPjfGJh6L4kKkyQ807HQ80Ec6sCAL6K0M3OTnJpRVZxQUqSa+DxV2/kfYzXkMklk38BtsL/GH4gPUHFPoGz+fhG/D3L4dfhvdnP5jE/jHedbs+z2+6+5phz6XT/LXn0nq3/9KKvf/zzjvf+f2Aw6dSj2OcNdwnzfo5i+iUrq4kDkOqf3AJtdsibcjbi+F3ZKaNjyf8LvHcdbG9gTBZ5O0BY2XZs9ek8Qq/LZsWu5c919suYBv8tT1r7G8WbnhxZsYsvzP/AYOxdjIKZW5kc3RyZWFtCmVuZG9iagoxMiAwIG9iago5MjcKZW5kb2JqCjEwIDAgb2JqClsgXQplbmRvYmoKMTcgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAyMSAwIFIgL0xlbmd0aDEgMjAgMCBSID4+CnN0cmVhbQp4nMS9CWAbxdU4Pmtbkm2FcARTiIm9CRAcojhxQkhCuHzIjsGxjY+ElHMtrWwRWRI67DjcVymEs0DLUe6jlBYKpQXaUsrZgyMBQltooVBKW3rR+9tQpR//N29mdmZXK9uh/X7/0lir3dmZN2/e/d6MegYGekiQnEfKyZLW9nbd6lz0e0L+cRYh2dUdvT39jTuvfoFow48TMnpvR/9gS9UPa88n2o9PI2SPp3r6Fy81Z2VThGg6IeS0yKiR3njYRbcS8uf9CfEdM2Ia0Y/e+3wNPHsC/h02Ajf811U8Bdf/gH8HjozmNrXVddxNtN/A+4F3R41NaW1HLzzS9oU//qQxalYtOewQ+H4eIX+6MZ3K5j5+nCyF10+C5zrRaFPS/Pg1v3nv1N2P+CcpL/8A7pAXH77vLfH5cfLjhvJ/VewNX8tJGWH/g/fKF3zcTEjFPvA8Xf4v7En5n7aDkF78Syq0+7Sr4d2KssvKToDvx7FP7RSyVFtGSFnQV15eVuYvK4dnAbWP1p61a7WniF6o4GPtTd6AaV5Ln5VpZd/EGcyAv/XaZnv8cxh0+Hc2fGPXZdDzDfy6nBxAbuPXFWQv8gy/9pEjyOv82k/21fbg1wES1hr4dSXZXRvl11WkSTuLXwfJftoD/HqGcr1b11bt+/x6JpnbfR2/3kMZdy+Es5xoFVXwbWf3N/m1Rg7teZhfl5GZPb/m1+Wkvefv/LqCLOldxa99ZFNvnF/7lfsBclnvL/l1Jak7fiO/riJDx1/Kr4Nk6fF/4NczlOvdbvxi3xx+PZO0/OFb/HoPcuAfCs98eemSJSsXNS3Ve418Qm8xMkkzqxvJqN46kolnc3EjqfdHRsaNTG5zSM8no2ZGNzdFEvlsfMzUE/GImcyaei5lN4qk9M5kpHFxa2p01MxE4kZCH5hIm416cyKhZ+LDI7msnjGzZmbMjDZ25I1MFIcwklm9NQXD9pnD+YSR4R9NjUuWNK2mRLRatKVNactFvElxH+vMTDaeSurs5d5+HT7wciSVi6SSY/RB4/IVq0eNjWYqF2tMxIeWwp2lTStXLZ9ylDhgByYwDLgxM2ZUz2WMqDlqZDbqqZjumvRiF1JcjxWMhzzQjavQYmZyMJU1RtYcMkdyufThixdH7G5yFLPw1dWx3p5K5vT+VCwHHZl6GPoZBIzrXXy5moczpjlqQhu9tyvc3B/W+8LNbfrAms5+va2ndXBtuHtAb23uC7cPdnVtQDjGTZg0DgxfciNGTp9I5fWNppkGdERS6Qk9lsrosXwmNwIDZcwYICcZgWXX9YERwNkkMDTAK/o8+/s8fV54sKs5BJ+8Lb1UHsM484p6mbeArUzCHAYkGHbnQ2Zu3DSTCC2dh1yREC5JSI8uHlpsuFcupBuL2YP+XKPekolHTYFZ4IHMhN4QSSUSZiQHXJCYCOnzXO8DOHS0IYoyyk/6UDwZjSeHAVVJoJjIZIA1MoxJBA2nxswM0DbFExDdaJaSWp6y3Yh7qQ1cHVOPmtn4cJK2E22yujk6lIrGTWyQMePJojk4upq3IARsPWzieo7HcyPQ94Q+agIThPR0Jp7MQU+jBsADcwb6NWlPmVQyHtGjqUieAm7kgAtBaKSj0AyaGNHoolSSXmRy46nMxhBQ1ZBORQGsJhM6dIwUDim6ZsQ2akwAxnQjm00BjunQCBKdXDqTiuYBocnUOCUNmBe9G8vn8hmzCJViQcxYjE1cbzAW6OMjfCHmNbe2hnsH9K7O1nA3sEVzR184TLkhhGTXMARUFmNLRpEdOSMf50g3IhEznfNYEWBepK40XfS2eDZCO2qLD8dzMLd1QFcpenej3tDaFtLb1rUB2ulIEQWqVNpkkwLY0yA7szB9Bd2xeMKk0x4fiUdGPCAABFCqM+JJkLh6J4M/mgKE5QCL2REqvQE3JsUdXOYcGAuJlul8JjICUiiEU81m6YPxZCJl0EWLLkbMZ2FOCQo+LiH0bXqTKUiFpkZ9MA24SRsTuDAAfyyfSITcjAgQQpfDGSPJJI4BwCQXSQ1E+SGeNIYABV1SGRUjAakIIKcLkYybWTZNKmgaqVxSOcWBuUR8NE7JDTrtBbWSSho4wU6KL3rdAmCA/M6yTpKJiUboI8MIM58BW83Msr6dSHXxmp4doZgboowbo+sElF6Sj51cWQw10kGIkU+e0sqwmTQzlAWZBAGsxowILCRcpQCz6RHKszgQMmGI0riRToNyp3il9EhtUDq0WFrRvHEaSGFTGzVBs5VENNUdsLoZPZLP5lKgTSf0tNJxXHQ8JDq2cYvwAjlQ5gOcUTkhiS9NiSNHl49JvBAHhtIzAoR9AylFTcFfgvaioIwz8aE8FWG0WZ6B7gA75LFEsRTHE5dxxVKXPgMpRykxlaSkjw/g9Zh8IqDAUVGm0QcKa/RyoUffyRoJuuhGLpsyQVFMjwBBao7Eh5C0neNlBUZC+tCEPk7xGQMEG6NpGIUiDjmCi/aBRdmReCaXpZiIZ6KL0jDrCT1r40dVCiZK+ZA+mh+mKiAZH+WdmLlII5OhWSHnhR0B45lUiyZS47bkovSST2bzSOmMEZB0KK5gjiirxLqHmOQDRBWZoxvoOzEjnqAdAXeLhnCZNc2NrI9sli8/RWfCiKCRZ+jZtBmJx2BSAou0FSoeh7KRxDZkMhoG6U0nlKIMBG1gyoBhEGMZekuMSGUZJze66hvBckBwkYhhTdC4YQKtmJPYmAJ/qDAoJ6B0iefANQXbjOoQoZ4ZcXCRwMwHb6mgSIFIxkQVDLDgTDIwg6gey6RGvUT9QAkYUdghEgDlggvphPPpYWpTZxXLIWMmcESq6uCr5OtR4C1YCDYVJriSE2oDME3jaHvADEQvDvsEdNFSB20ZgqOBIrL5RM5IRiao7QKrlItnqQ1ngECN4Ny5UqQ3ce2FPxTjEoNhCIgDbHcjEaO4zuYj1JSiGjeem2AiBEah3GDq1LjOUys5O5LKJ4A8EtmUrXmpdcv7B+mcScHK0ndijDzjTEyCgxYV1FO0GIgg1rZUE6bzNoJByJZlLJUYM5lccHAWdwWdFM8WWwCJStuWvQinLdTp8NEU79TclEOvIsXGAaTlTBCOGaoG4CKRmjDNLJODUTNNpwgAC0M6lcnqeb4CRdZwVpktFS2MAuhIgL+xOFrkcBUzkdEBLtsYpc6Mre0cTSiMYDPg9HExI/EM0FOWEgoqJAFzRhhG3kuMANL1zZhoRiICpr3W+SxQIdDusmnaG8W6QxgdCAPVhkDkoylACJsz9JBEQc2gBbs/h1aWh52vCoSSKp5Zp3EKXQwdZzoXh5bldhuzVhIwnSFAgd5AZwcyGcx2QPdQ1syJ/p0NwW4GImtPGNSgpYaj0EJyECZ5zU1c8eELw8YoJwKKE1z3rATZzDDDACiU0hoqA6/J8RUQ4rJI53O1xrsxo6ttkwIgSKWiWanIjSw61PZqCfcGegEzZ8RIDjMyAC1kgrpFqjBNaZlRqQVeRYYLtzWpcWoU0HW0l0tFSBzlXNJkk2N+njo8AJzM8qk7fAzHOHRd80khX+lyoZSTQzbEEPNoRUj3URGqFEDbukuYuRzTiTTUEjHjaWbp2NxBFwEwgaqS4Y8NCLopkULKZW4U3FDtVemvunAQiVOZQqebEqxkD8T6p8vDZUPIqdgFIlCbF7MZ8OihYN4s0MMOI3XIVE0wYZ4KHSTFLPXNbI6DUScUyGG8+QcsbVqyekojnDZrWq0sB1oP8ayDLATfG0nPiVDIbe9kKDVmLtAb4gt03hPaqHqW6tSILswHkHhArc61FzAPx2MMqgW2hUv5xoxGnd6tAI+p8IZ4fIEUJS5CNiiRjKJZZAhVlUWnDBd1LG6Oo/qGfpAC6ZcG+o2uJLI5CDlGiwlKJ/TKTAI+IngJYKK5MUHJy1blnlOmExZzEROYvq9EwVENopIydUi4SVnu32ZM7tvA4BlUBGgax0cxbgMUA2ZF1hxJJaIh9sTI50ZSmfhminOhaXGm7qfU5sllWRQX5233JEYLKboarXl8Qxp7ABiYakxLIJ1nBbexmBZ9B3gpGpdm6YgaJqTPucZiBnCK9jpE43G0y3iOLipTFqWMm0Z9aMEk3n5Sz3a29+kN/cBoqOs6k6NAldiqz0yD/Y+QYKTR5lBFxKKeZ7235oG6m5Y3rUbv4RjqFS5CC9uW05RUcZY0BsaIhdIjPKc36AvUUhdjAo4iI8lUIjWM1mwD4yE3ZPqA0srmLsnW6WjMFonoIJubMJ6UmPBwlU0DVpnbJkVCUZon3FoHMbe8sTiKShkQyIUGjKihxrw8VJIhZXDwigzKvyJDwPyQeI7FtQxmG3rEdIoCIcXhkmJdXBzUtEP4WRYvcepEfI+2bSwW4dPDJFI70iwKmhSLl3taEcySYj6VgfFJGkQAm9iggOXTDB6veXITxEXvpd1Gjy5E8NhlRNs2kIi5UXvUpI5+ccaD98HHyCcVCaIw5TTBx4g+NTtiVBYzCi8e8uCsAhpPMKEsMZhbQIdCHxHeBvsWSA3cihEjM+pISTE/RKyhy3WhsIicFV2/omBCMUgKIBnq5MYxsMIQOcYn4p6bLSpBYeQTbPnhnaTBQtm6fphKglQIeVsQlDhFMNZgeZdSgWmKIeAyDCHj/Jjbw4KZ8nkiFaFZFSBOHRCIca2GeV3N3fMWMPWKsUY6BBeuqCGK1lLil4W6omp0lb6dzFNNgnSRyjFXK4NaHqw6Sv/sFZQ73g4oCJxGs5FJYtaF7JJ1Nj5C2Y/61p6R6ZBtC5R+fRRF1IgxZvJQuLd0ovqYZjOS0Fd8lK7fiv9IgtAXKHBom6CsKClFJvVFeDARNFuSCkElbthnMvsp5+1NY3ZAKi94z45z0LAmPqCCwcD3sxNZ8OKzznWXBOsK/zNuMKJM+wPiqU+D2kK6OeMZipSkktxzCx9AEYY4wKHFRtEQ9/O96B8VDhphOVN46ioAtDGfZ8kwIvfEzsinRDCpmVuO6H3QZKnApVTxQ1QpgiZFVxCTvpFsyViMN+6KY2qCOoZMpsOolmG6RHWzmT3p5cJPkj/iYWka2xWx6hCCJixcNL05PaUynka407IW4TpmXONaZqV9LcJ5GAgawbxzPB3nVIKSRZrUGxTG2JhMUfs+McGcVhi35Gw9tQ2gTMxzgrs1GFBjoSc1LDZvxIhsDM3DWfEp2ZMJOfNfzGPwxioohtSYF2liKJXCIieqr5STpXULKFcamhaAZRDZuCifZtLAnaFg9AlmG1gTCSVAlKT5Xspe0jTJpYQZJrUpZmOiE8I0ByyC7zJmMnpgsS8YlbMyz8SpQRJFLqMEg+6oXIJ+GEbAMhiOU2YTqV/8BK3ForYlbPgBt2UBepM5awnGbzRenGTkRkMqckU8DBKguXhWMU0c6MSKAjsRPZRBk7gousbqI2iaEjg1n8gJPS+9riLUwOs8WwkruwrDz8mULUKxnbkJmDYrRX+OLjp46zRzjOCLBS8mHxQATDJRYo5HaAGL4kjDE+cLDk5OenKMcPOz+CXljlFhdkMJkuhDYLIYeTUQznL1kzUESZfGlWQKTAzNRbhiV5buxk5ucI84LqwqQXbTpDJmatP0ScIUy0blGDV9GvXuVI46BDTky+hNYSMBI5dBPINEPaZ5CPZiAJBGIubpvW3tivxlydlYPJPN8QgFraGhUaVRcFdSUQSTyzWmYFVJmGVr4sIG4hRRRTPuSxjrFBkGJcz+knddLpNSk2G7Sp/IaaBhbQYcRhNY9FXmodnYGewJvGEeb2ahyPSE9BsZF9FZY3FLnsVQ2dgJY9xe78Ekmp79OZZi8uyLOyriNdZ1Uqgvg0Ue5IAZ9gjDpqB74zRR0IwZ3TQrOeBlU8hzHuYerCHKdhDU2bzBGM0ZBEERyqs58FEWjAwJdEguTEhBBksgMM+HhmEiAhjArmlggkpWzhRVajQxzeP0YNG2CKEzn81xL51aUNTr4sl0RAEjfqdeVwwth+2cLZJMIR4HQyYD+0UmoBFbIT4IDepJy23cRN9rtedDTFhSnobbKGKxKEsYY/CSHeKb6jEaiNm0ERExxY2w/nC9mtnmSQxt83aMPVWDM8orgJixFOL90jyjkWbLmMwmEMUZapxT/yAJutKklkaUFjGl41SzgWUHSwCsjKFVZog47A+w5s3RNK4Z9aNQeyPdpPKZCKWcaKl8DbBxCjDOJFJqiFIcnY6wwpX8OJorRT5xp+rtSmOQvQLibhikAiZ9gCNCDrnspBa8a5OL8JFpGZcSEgRTyuaIFJcV05c9qnuVNWkVErUknIh02JTTo+ci0xKLJJiB2aj3IGuengcPIBqPsJcpEMJjiTn9ERZQYFztqs8R4LsphRE7pxXPCAAzZ3mGy16nbD4Gk4szt4ARKF8z9D/HjDgLj8aciUeKXclySSYR0SGkje2sGsxvOGOM6g2UnxEEzH1RbSY6Qsn1iYCzbYWhiaKVRpUuyJcGNsEkWMCCQIo6s4MwIhRIi8YyUWra84ghVvo5Kn/TmByMqOJKx4ySTSSCx5Rq4XFQcwcD1AaNDShBaW/VWqLWK54ETZ9FmpfoGjKTZizuFSujQaSmpfZwk5cNi0rKVCKBjlVRopu5zIwQGf3CKnT3DCgarZQUoR2riW0W1aFPZDyZJs3mH7BqdXNX75rmlvCA3tOnd4UHBsJ97T19axf1hbuaB8Jtem9fT9tg60A/tqX/4KneF+5v7go739Bb+8LNA5093fIVeN4WXtfZGu5fTVMD9O3w2pZwW1tnd4fe2a2Hu8KtA3093Z2tsl1zV5fe0byWthjsD/frzd1ti9V+Ih799Pe0D6xv7guv1qPsaf9ATx99xl/uD/eto18H1oT19p7uAfsFnU6G3u0d7OvtocP1tOtrm4/zbiww0LyuubOruQUQsK6zGdt1dgMSugGHg/30VXytL9zb1dyK5a36QLh1TXdPV09HZ7ifoj6+QO/CJCzlMCpHmhPpEWPIzIlsoFomlpeZJ1U3U9OJJsM4oWQjGSM9lEptxFVmhWa0uoR+pWY7ditSP7RaGhiZGm8JGw5aEZS16VK1L8RQMAyL2Kk5fVCh+SEaxQPKHU3TwtW4uSiSzykNsiASIvGEcgeswlwKJhGjBQUslxwdMVGb0IDNRuoEC5SoxQOygFkk7rm1xoTihEgQxjxLpybLX3iZZ5g/wrAN7ZoGKDCVyN+3AYwYtALJxjKNIOk0kRqWaSe1wBTjOQ5nRcR0nImvIjfIkcbiHrcz25YVoRcQCgma9xQRJTssyZwdNikRPcpNpwRDTS14hT1LxdKaRSyDgsrzXWJIbO9l8pQwJLD8NGPbEtPOnuncaUTbhdppGdvrLJbgcbp0DusKQ0WA9iFDrdGl8TqWZUS1S796F4UpaySqVgFbci8HDR5GEFciAiVsfLuQP5oxqIqwtYKExl3gglU3idRwii9ViIWdgPeMyEZjmBMPtEmn4jzkZyR4BRraBEYsZ2amRikgqpO6NZgkSHHVU9xMiUyVivEazNI4pnjbDU5qNIW60jZFplWoWircKypLsdJj8mK7SWpb9aZljZy54eWwDOe0sSJcIbuzasraIwwp5FlyEaVI4AAgzXTCEEWNrlAR9k0Zm5bDOHNHDC/tjjoQXnchi4zVwmJkyOLqYVkWoW+Mp7JAeLx+y64uBr7PpGg5zgSLDqHhMmpERlDYDFNvmc6Akxl8H0o43o9ngN7TfG8ISle2csidI8wOEnlazqHOMBIX4ZOL7bijiFZZgKwadGfZT6EvlM5pn/FYDGYfjzkrCowcmniHFqUkvao8eSJNidHZpc1w25X6mmybRJIxAdYY0Lj0sGlXEIvVp2AOTWCdIaKKJxspKk1ayJnQW3sHbeIRprDY7IR8IzhXVUJqcNBkJQtYrkgjXynqNPCq9ow640Z7q4goNbZLsrhEiXr6tJNIAspzy6eb3hMgU8SrGT7ZxBFrySCzJ0zcfJPFYkfcFYceNHNV0UARAlvJ9pSoCMB4MvVnY2jGJBh+ecraS9YAIXKjBsuNsL4J65ddwVlWEGbanZsmd5dESVCGVge5KncoAHK1vcqAnGLPrgPDfUwirjt5PFf1zJSm2VwqI3l01AQynmDaZMTIRFn1KEoGur6HFTMVtKCVAlm9RDU+ZgXS3Iqm1atgKlGSTEx4eZbU1HHUpiOFgOVoAqIbVi1ZoEeNiazikEXNBN084J0GaNQHeDjREHBO6JGEER9VdJ4Z5zo+l894CSruUdpcjGzBjDPcmTbEtpXROLurrsbOTyBtUR2etWs5UxlWqh/hhc+gYKgxKDUaWojspqAw3m8J9z9jDtP9uYATVrhGe87GS1uLtNQaZQgP+ibBQR11eOpyh2AUs6jutI17IynbM+2iD156Ez6hFxzS/q4Neltnf2tXc+fafnQjwWnra+4eAJ8rJNpQj1DvXNvb1Rlug467W7sGqRMZ0lsGB9C37upc20nd3oGeEPPsWFulL3QRw32ta+Brc0tnV+fABuy1vXOgm45AXcpmvbe5b6CzdbCruU/4lkDePWvpi53NXfrAht6w3tYDvdFBeecwYPMAjtrTG+5jvjQMVuyIru+E2bWE9cFudDz7BnspyDAuXPf0LWrvC4dDOnq2vD/n65390MPAmh6Yclu4HVxw8EkHu9vCfQCM3trZ1zq4tn+guRvcbb1/DUWkG3AYuqsTHeCBHn1DzyAdq7l7g94DY/Xh1DeE9PVrwvgV/PNWGL6vuRVDDOCZD+gNNub17nBHV2dHGEZbQB9jF+s7+2EC7bzX/t4wHTpEu+kPHz8IDjV+7aF9t3a2wXcArK15bXMHXWnZdVdP/wANRsDKwP3+ZhoDwKBEC3XT6VrZ6KOobgbqoJGNwa4BgfbB/rD3CoTXhbv1zna6ep3tlD5gls1t6yjS6Av9g61rdFj0/k5GIKUqxZCbqKjH3VwioMpksFKllIgbQ/EE1Qd2wdOEjKS5qnxssx1czIjkLCeD2nzGNlokU67oE+vczYQHo/tTXI0l4WPxZtwxSGte47ZxlzFj+aTqC0dSWW+HybYeMIUviy892/IRbD/dRopRZLcfnHX7f1TrrGjkVNvVvF5fhCQ2CLQOZNLdBcvWT+PgziAy37BrUC2hJuGY3pZWNouziNJ3FL5ZGhVlRhO1nqktSPdCGtQGpcYxmO7UNOTqJM535RTfZRtv1P02TAGwfXZZ0HDUoKS84xJfG2wW0/s6O9YMsJgdPG/ZgAjoZPPWw+1ULHSukwEykDRheB3VZFvzBtkhSKrOnrZG2gj4gvIGCBFVWvZRpLZ1UhZD2UnBOq4TBGZzOzDfNDt3BKrVnXxDmHUsShFTKvBCXMY2+4ZUHWQ7mmhj0zwCNQmWcZMAbX5eIOI2Irll4rExccjkJwvITKqaH8XEKP3SbY5Dz9TLH2zsb2xuDNnLS2O/E2LzJI3OZNlGcjSpMOdBIzkp7stSba0EZFjVBAsPc58Lt5zlcLt6DKxcZCkECW2ArEi2CJVdsklMCcOpvOHGEFhkaAJwuSTK89VXbMcsx8vE9JgZxW1HERCNObuo0wtb/c0hWVyD+zxZpMaMUfFBF5vtJ6QYEMOgz8vHdcKR0cfMZJ7DwE19u3eR+Ie3ABFYlEi9LSOecEj2EsRhqlF8V368m4f2W2lGMymjUmLBRRKo05H/7qcxG2jXgbEfR+wNpVEUY5BNq3hqOUJrtxJmdFiqHVZgCU4yK1BArwQjjcWuCXc+0JMWyQXbkXZUODk3DGQ9XQ4GkgvDvJaEV2nwuhB3mhFpRFUFxceQlMo+cuWQpzRoRjG7gr5VOpVlspjtbTCkF5TKsAc8OBlyHZ9BXdR80k7MoCBG3DBaB8o/nTIcrfbkFbeunZzdKX3MoMkZNXsl/HHXEvA4UTSOW/twtwJtiN4PFWPUUI+YmAbk7GIMZenxMLzu0x1oxcJJu1YJ9RLW9koXzm5qv8q3UXPEo+CkwsYbYozRcRQzqeExK7rByASHKcPko8E4nkV8RtMmSnoHizICiyfHjEQcXPKxVBxzyPmkgoAQNzZGeUrAKbY85TQVrnHK+fyMDB07Y7SH8wzZxTp8bGVatqjnlopnZpR7bPItpHfWeZZ3nFPoumgYuol6SbF7rO7wwIAP64zV11D9MMrT7LxKkrJk0ck+WVFJTauoHVqD8QTbVyRijPrSJlbm5MaikacHS+R4aFVU5NkEyQLp6NMyYnYWCmZ49R/nTPo6J00mNXS6LxXWBfDI1FtRdsR+r0QhYPHJJx7YU6skGJ977TPgp3SBC5zDKCjoRzdOMFFBo+lDvEQ1Lup4KWOJbROc8JMqMlLU/2acVhx3XrpUrRVi/EB1O90iKSoGUHjGlXw6zIVq8BB9atLwJlZe0WN66MY6t+Sy61DpqRTpVEamlOyIFsWs2Ag14QjHu/frC2KjcDFbCLGTz7KiUYZIQ30kLJ2QU7HEjFHwLwyKIvSZQI/HxAajKI/YsNWT9WyZfMIUKzeMnpSXpPZOcvBNbll3tZ7BCt54foYm1mBVaIVDPEljLXGxN48anMOcvWIpEQBSTlDg8Uss9xyLG7q5iBoT7EyAEhFQSmFFksUcysZzLN5MY0i03DCnj4+PN3pkU/SxJauWrGxaXvKYs8V8+uSr5MtkKVkC/60ki0gTXOuklxgkTxJw1QJXGZIkJsnCNwOuovDZSkbgbhzu5eAvvauTfhKBu+PYPkc2kxDcy2N7E+7o8HcTtEjAvSy8MwbfdfgWh3smtMri9xxJefQUwbud0CpCGsliGD1FRuE/2m8Ex6eQDpAJkoZ7jXDdDHfoPQrjMPSVQ+gzOAv6Fh09Ci07ABo6SlSZhYHQ6DgKm20ftB5GfNC2zm9N0AvFXRNZDW/0kLXw3+qifkWvos9Frl52tf06nAXFYwqhVqHohbd1/k3eHYGWOcRkEmYv3mgky8kKeDoKvW6EPmmbGNyl6zIElMDaLIXWK8kqaFuMrzinC4bdYU4TJn6L4opm4CmlgVF8cyPcS8EY+hSruHgKKpj8bW/6DU2TciWdt2D/OY7lNXCf0s8Q/KM0lYOxDgdIFyNsbmhyNj2yp5NDrJN2HCWH8FAM5ThE9FmYwzPI6VcnXS7OaQbc07YmjsD6oXzcBe82Q49hpGN63YajrgFMUjppA5pthX7XwrNuuE8x1Iwt2+FuF/y3QcHHOI5Fx5EzTvJ1HoFWdNQJeJKHz40ITZpTB22fhmc6zCyFM4hBqwy+x2ZEe41xykni3BpxDhTaEU5nnwwPDXwUnczzeD4P74dxts1AI+y7s19x1/ttNh/3O16wzCMLHDyTQJ5hlGB4QE4pLYdYN5ECBW7FenjxSEjhEnodBfocgn/GlDwXwp4XO97oh34bkRMo10RxXk6aZTI+g2vbgBAkcF4R5DAqaRPwLIQYmnx8hh0xtyGbykwbX0PwThKlTxKwpXN5xmRM5BNjrNFBY14UNAytxhBqphsEPTFJN4r3mFTL25psZEquNhTeMXGlTJTpwwg568/dTxY16SjMM4V4MJUeaK9xfHeqdSgNFV2DENfEw4hLwZ/j0Av9xuCmqz2Ko9NVpG+kkUKSiJMoPjU4ftg6M/lr2jBlEAbKMXTmKfjM2xg3EGb6nFkRaWjBestyKo2iVkzherA7GZxFCnVMiMuqIfgUGp/xpmrFiHmklFm6oVYlG302wWlMR12QRbjj/J2oA0ti5dI40yjMglFoEr6N21KDrZdoS2ViDuWiOQ2qdHMIlZ8xx4pTSjCQq8ZxBJUjqDRrhf/CoCUGUJJ24rduri2aQdtTPRC2dUNIkXYNMDqTZTEHlwnKjpAz4E7cRen0fgS1Qm6aPMI0r5RdaZvT2xAzERuiNuScOFIPXbd1XF6l7LYbEe5W+BbCe+vgH6N2MadICVylUD6pKxXh0DCbMstX35u6Y/CZwLfYao/jqkYcVDIZDhgFCFlnYC/MfqWSXsV/FGebRIpmtJjlo6QQuiSnbZ1zaopDUIrGQkV9ppE6I8gTWVtnRDhvZe03xuGNBHwaNqdRrSJpPsvXKWFjX3Ihg9vcJWnKbIUm/BwEKBndpJFjJccw/MfQNkxwe3Byq2wc1y6BOiCD81BtHINjJgmyyMu/EfqBjmsAxzAqcNsJch2mogQpixjOBUckURMIzSRsJWbRMIxkOTd56ZTSNEf9gFHkKaFnKKS9hHkfKZyVXMFOm77E/RaOjSSnDhWSJOqjRg5HxiExKY3RNczyWUm4J6PUyfWajrwgaG7I1rgxm5+YTN91fTyZrpwOrqU8CDmkT96WK8N4x0QKzCkrzWiC0WoMacLk2nsC8cdoNs37d85IasKQLcdp2zT3ygW9CvlI11RgO+vBte7eG/9LlKKuGh2f+Wy7TtHC72C8m0GJmkc5lOK+6QTSnjfE8SKIh4ogLqZbiV8mHYTmY3Qm7AkvyZe2JUfO5j7Vxgu5MCPks8SQhJtJpahDknp5rDpac1nUXnGcn7DCRG95B9ZLYzs0TS6KKVzERnDacdOxdcV7zJYTMjGF65JU7IwM2gIxhRKc77hxIecq7TTxhrfW6HVZemKcLD4XnG7gfFPwnXkU/00JyGzNEVw9KbUnm1+2iEZCSNsTCK2gzxinYANxy+YiKE7qCKfVPgA6MYuwZHDGIRv+DNruab7WE4ihYvop5SmYii0fQorPAy6EF5BEaJyQUMkcQTxLO1TlvyEibG81HsHmZ9q+aAKtmmKbS8iXPEq9vCLTVY0gpY6gK7aO0q5y83uIqDYfo6ipI58b7HFiaCkmbIiY7nb3yO5mkZI2OuDIKvFFJ3Um8G0ZyTOwhzSuVhz1acSDFkVf0uMp7dl4SbYhIixoJoeZ7S1WKGVrIJP7pQyuccTCGF8F1so9R2GXOaWb4PWNPOYgsSslMeMTGblRLbTp6CR1nm76kx6G0AnSdqH4HEVtzeJmwg9xe8+q5HBaCWr0YVdsBW9bIILjGwq/TDjWJMPXIIrzzODMpmvVD+wiHqVlJymBUblbF4oVpvEFNvuoLVWKYw4ZlAFyjsKry3LJW6yvR7neinF7Sq6KanEleVTGqwcWNY0TU5FvUQ9YSsdPmF+0dBK5ZRC3jo5wKPMYxTYQaxN23IXxUg6pUsThDG6hRpR1d3qKoqXke3fuR2hLwYUqDTHJweLudJ1jNl1nUduKqJTwceNI1aoVwuYidIPJW0bt8TKcilLYMsq5LYt04/Z5RezWCT+znTO4Yhk7JiTnxSCJO6xJlh2LErfsmZozJAWp/e5qL6qft5GwCKHKLWOERhHHFOk4uc5yZvwmk/EqZ7sxKT3tYrtX4rPYUhezj6J1pUJKbReBbzE/OZ8hPmuTW44Z2xswub2Tgm8mEZaYWMEoSl6xigzD7oh0CmHUbV9heJrronqizrUVVosqA8ScGP2NYeRLjMbuxYipaHSGr+LIqMjMFPt2pXsReGRxBrn6kjMjOAMmn7K2RJEekhvPwjaXEaNd4WKJQcG/9LuMRkYVbP23+TqPtv4wEfGoZf/l+MZ0/A53pEPiQfiGTJKP4hoOu9aZwZBULGoVtyzenyMyljW9eH4pC2HXvXg1dhq3cSfyhlFlXUr7ss54mxpbSfDVGeJUQGPCYu2Yncyi7Yy6hxD6XBH8k/W4gEvuJM4zgZTH4kQi4uj2hbxmotq8VLY5PT45wjCus1MSCDqR/F484rgtc+WsmAwVck16BtNdOScPuK3Lqf18p7fmhIa+sdoeQ2pBhoMU9pslXh65QURu0dtLdmdv4lxebeJyI4mrrcYUmFxl3q2UFaaCA5XfhK0V53LTabmtISnMYzGbS/BjMXeVopC4Ys8lielYOTWfV2r2Oa6Js65VL53HKD0fwa/UQnDbr4K7pC3nNcsGG8vOWIRX9tHbUhUYLI7dJZCTRXZEriWrdaA0yCJ30vpw6w7BCYwmpFep0p86Q+Y3JTg1MPjUbJTBOdQ7vuqVX52cDiJ8TZh/kCiSg6VmpMIvuMdpN5SKRQ55UoT0zaejzZgePZRHb2h2LjxJJHWIyIyrOwrmjp66/SAva1bkzYp1HJvrRAmcs/nNJwdg7dQSrMr6TyPhorcmlHRe3CFjD3FFexe3dOt7ZoNNd0UEzotzJ0OEVUksQGji+OmEScZRdW4J5riedUcfmI3HZOtkfO/G8zB6RSquFhDVZlctziGEXHhg07NkVC+czjFuZ4ydVsnkEtmwJckokdEio8irYjyXtP0yxqljqCPHFe+bwSNloHjSYD8TPCm1ObPkVLmYsOWJuEfnPMLtdXmXYVNGNyZs6VXslU9/lcUKu9fFvQL/F3klgZ1SEaJdt1OHirJJwp8T+VvWTs3bsJlnFI9ARo3jnC+jigfCohUUMnqV4JaffMdAuhtB3MTJZpvO3T6tXNOp3hVxnhz390SdrVzvYpjccwt5wqDG5uUYXpE9hjEWVVN9CSnPRU7Wu05LjMP0kuACd7SUvecVmxbvO30sNQKcsmEdIqI+TkAZ5zyewTGlZ7GrkZtG7HHBJ6JNJoOywC3tpA+lRD/XaNKvo5w0ymWl7KsP9S6L/0ucyJrGYh3qbcVKf16FvRW+M9ndRJaj7Ja5h2PsXOEiJYZdbE8LqSrXUtSBqZJFyEf2vmgRs+eZ8ZwnoyNK50mk7RRqG8FBDQ49NBXOaPTLu69i3eWlrWl0OkaKrUSZQaa6UtYnJRR5NHlW2URIR4g7bjK1pegVPXHG1pk1txw/p65FFRqQSRdRYSQiamouT3qSoRIzZ7kiw9a/7v0Baj6EjqvWa8kn06/TmboiZDrVJdPxi6dTqVlchS9iklP7iXI80S9dvamt8P8mTUrZLuWstGhEXjPtsJKmjkWoMSk1T8XsFRb1FpUILE5s2BjL2zVmu7aezijI5PL9k2QbpweFu/J48kh0cRzIXecm4qPM35Y1A1NV16lwOOeRRwvOywbx1pT/XezLGn0R7YjZdrEqw6czy4N5fKsYa87dSdIuMRzZAjErmUdMcouR1bjkiMhWjODV6CQ7oNR8iJsPJ8+6CLy491AJ/pu6MmE6WPLGSIaYtu0rK1ZUihxzrchU61ZsVTIPI8+jUmKF2DhJvCersul/h5WUgsIS2pUYhJCc7spYgQFh4+1axbSgIabLZBWyXD8126NWZnq9n0A/TOxVYZJT5xQo67UacB9MM+nmuzmk9yrrGsUsnJar9CGm5ksv+lWruqKkVO2qGDuJ6z1ERLwvxluI2JjIeUhfnsXqhPxXR5H2zq5kQJmF04g0pdrEKhReUKqQjaN0ZNpP5K2nXzMdIsVxgU8y+qhiRVG+GuOyXdpsu2I7Cf9Y7M1IcrjiiDnGfyv+f7RBxAgCczJuIu2KXbdFPnlexFmZyHy2JBGWoHe9IfVJ1PiTjBhN139k6+nleRlcervrOUS1pnxDWAyGMn4WZsF2kI4q0tR7zYol7OTV/6puMLgfKOxcg9dgmQ67xDubM446hVEK66s4SjCV5cOoSFZxsAyt7Cmq1HOyfP505b/0cGQkLKfMSuTUS2FA9Oxcz12vRnTmxKgfySSLmt9qdsUcZe5D7Cx106WXFz9ke4oRbgOJrKDc6RvxiO58Mj1Tiu6mU6fmlh1MSkk/TPgyql9SKputxienm4X/ZPuPnNXSom7XXVcdUrDmjuHKqLdTPqn1JlNHwieLWbur69TIteTLrGf82l2dJyuCGG5YDjDOK/FUWSJtFq8otfDf3Ku+kbA9gMK6kh46HX2Uxwh3dW2n79vkuKxxrueEA/dqhZpa9VSqWmwevhnBvZfzlLVyrlLxysj18NLVao5hV2iVeQwpexfWdHaHiKpUgRevFdXJSs+VFecnSHulgTSh/TvEsbIIIwaqbTDVHgpVfrJoG4tNiIoSdwURs7PV/JRX1IStljMa5uWbyr0xUR51UKPmjBZZ3kX4EQJvat0Xm6tTKzv3xJWqJPG2l6UNxqAT9hKDR6WRFPcl47Zmc+/6ld+Zr6XW2u5aHF5I/tIxiyzXS1JyJRz6TdQXy9wdy6kbRN2B5bSephMhYXIujmviFTUpTZ3yjILiHdFDXD+xVZ66dk3iWu6mZDqV1fO5/XmvXNfUVMNGd+6tZDy7Sql+ThbZrSn0H0R/lBeYps16Wv05m9NZbl3sOZbYd3P4dKSPtABUm0lI5jhyIzv5xTsjzd6ZbITSOpnNeno6xp3NzypPUmSqOiq5d8O7koTV5TDaypNSFeEyC/HJe2Q2XVrhSdUDc8/aaYV7xys/CTTFOzeyLhy7Y1VuaffflWVqVFvsPmF1UU5uE/aYiPpQ7upGnhIZAlHlq8o3b23kxqPTDnLuQRI5pnkKthdzDIqaCGp79JI2mKW3/avunKWWX4brEbWGQpxDI2qVRonJpWxUwabTXlM92FI2oYyDCL+yNG1IOpVUJfa4L3FonakjBrsW7d/1tpNnmbzPySjOKv2/yzSIam0Vc7I2Qa199doPrc47o8CU4xJUrUlQqyKpTvDKN6q6SKy1PLklr8hP57ypPhj34O9BlN8Cjn6SI+oupunD5cyouEdToVY1snOfktjDUzzDjOMtWW1q8rtiR0EzkXt0pUSTcsNdxTfd6B7jQ2m3M4s6i/hWNdpklSDSCnWezSHfohbcaAlMhzw5RsVe8f4LJg/UnI+ohokUYYbRrsmtBnUfpxPGqc/UaHL4PJPlYGXcImRzfpzjSM2lixiUyHWpXB9RqECV/JP5694RrdJxZ/nOVDusZD2Y1GQs/uK1A1rSVsg1E1Gp5xVzG0eJJPJeq3fhTbnDUuhp1lpasfKkLHdkjI1UXMX3n74tI4hZwk7rcdcpbuT8z+7Lun1xOqGo2nb2p2rPUhFOaY/LaKaILIVc8Ir9jAba/5Ibkyg3JBUz64LNmUmYJPcrTSJiGlFi2hZT3PbZWMyOcQHTyrJqVY2IlI5/GFxijRJxYlKKyHyU9L2lvEnx2hwhc1jceFf218S5rcNoXLWRUjADIePE6rhj4d77x2V0Zeo8cadDOqm5Xa/IoDoKs+6Gua0gd/owHSEkq5e9PJlskW2LpYs7jyxO4/KuEsxwOe/WESmXXfF/YfeUyl5lubWhxiQmo8jSccr/pnyeOmrJPP4Mt9lFBJPST4+iNU9HbqCwRxEqdWSBCXeOJVYCWqOonkLV1ZOfn+PG/lQyRZXsTrky/RoANTqrWtQ5D37KYuUlWzkZ42b1E1KCOvlM5j/HCDtpQlrIzr2R7niloF0vLScymkw6ygyh6Ll4rxpbP1a9QiVMg62fJRbkvi/hm7khkjbX/zvMFccVmL08FU9LL90tfUXFJosSLHBUAnl7Z8WVMO6qQHHSWAZ1ZZLrFtVrk2f6lT7zN03kzkER9y+2rnQi9ygVSxK3HvM+W3ice3MHc1wbdt2Ad6X0rnitu3auF51TnvsiUs57UdcQYTtGY1ymTke2i0qkJn66g3N2/8lpw079FCPsJFGZsZp6R7eaZVYloip/GS90g7weIN4+2q7aIgLiUju21Vod8Y5XfTLtgeZuaCX2KrBPm0kX6SVr4LOFhBHaHqyb78JvA/C3D/BJ763Fk7rDWAFF77cRGo3qg2dtsD6tcK9f6Vd8tvP+6Jv9OFp40jHo6rKznAfAWuoBHHqNwt5vg3br+Mme/WhrDznmRk/3bEFI26BVN+kgOn7qOIsw9taHY9A+vPprxlOidXizGfoSfQziKaL9+Lwb3lpcEp7INOHphx7aAZr1/HxqtgtOfbcfntJRxHvOkfsRh+vsp/QM7DByQg8/+do9gm6vjGjbC/Pqg7899ux6MNK4Ft44bpd6dtNAM0DWDD10IZUxCqCwNiv9dWJvdBbdnA4plsWocrQ+PNO1C893DSsnew/geq5BnuuCfx3wLp2FoHq2l62LyJ2wQocJe4TqkjRqDHbOsntvYKnTxPLcUnDueSrlN4uok9gZ5pQoWbQbKI/TKGeKn880TORJFM6zS8RTEW2X0Lp3/bAzPeJcI4vIW8IDH+KMoKyHvCwVv3DPKsVjZ7LGrtQ+feaF0jMERC0ek7nUKszanqYJsifCMe3VQ5ZbCRG01bzb5LjHmeIrESPihALpp1JJOUJMxTcRFTYbicgEu6mk1MkDXicwu3fcO2NrqqXI8pTqDsIYUS3nyU+d+qT7L6YbPZP7j2S1jYBaVFDIXYnO8YsxGOG2MbPR3LQsapAodHG+S9prt1OpE0xlfU7pzIq7TmeyHV9TZ4NK78Zy5rgn29sm6Uhm/Sn9J4jY7+muUSqullQzO+pKuWuPnH7Af3IKRqldC9Ot9tzVujQRQZe+j8Cqc3+Xe5ay/+lGeXYtIiFPP82Q4rjEf3/vmagsLo67iHgaq4rZldlInmNcVzp2JauKGLUPEcNBwSrVi/o6dS+j9HbF0105Kcybj9xnrTLa8vpdDlF5GFHoyl0D5Y7jF5/IH8W1EF5Esa/ghRuKz8lOcBF+JNufMExSLq4SEdwcj1HoPCa9EbWGU/LEOW2nuASUmlucFiMtDBknMBDbzjjsf7pDspOIbI3cScDqVqTXM53evGumdrWOV1YEUuwf41nh4f61G7lSo/hpklJREVmb85+eqLqr1b3uM0vlmR7/ycl2u14pLfJsy/iuBnelTtizOqeNqCfhuu1uVZrvOn7cJyTTlotsGcl0AJOabGez+6TGyauKTKXqgWlscTqMqIzw2nek0os4Uc3rPBDneRdeJxmXOrFYasjpnD3sdVoEOy2WxhSZxHOev1V8djHT9xmUOIy2JCazRD3jY5Swejtp2QwTkVsWa+CUZuz5ELeGvMen2pXJ9zRx/m6ItF1VnpO6k9pM464RS3k+k1UjOa3w/8TaFr+SIXPQMqLqzQHFsWT15CURi3b7F96QCzjjGA1max9X/JFSO2VkFO/QaeySnO5Zns4dad51dMWnNpuccybb9VU6jjp1RFFqAhY/TxJZL00xXHwGsZv3BTaHbOvdcHjwzp2NgiqZXBan8bViFKdY8rijwiZxniMh9Y1b55byhEpVDjLbUNTPytMVRc1XiohMg/OsdmcFvrrGdFXcvyriPtW4+JQsp43i9JL/L2wCoeeW/9d377mxLCi+1B4+r15K17WIymyRI5W/fMNOgXSuhuD9MX4tsqoyguK2sL339kyvPsdthcv8bIzIaEzCQb/OXdbTtWuYRHRGaphMUDlTPX958spZ9YQw0wNysRNPZpfcpwSxNjKLXOrMHYEBL96e7mlAk1l7xeeByd9jctfr/if1uaVyZt69st0boobVqUfZiUwp294Qvgmr8ok6zh6VNoPg38OmpalYH+JMAWdt2nQrURLcY3PGosXZqzned1yx3KeXsxRRndLnpksZkuQ2F6PoBrKKLME1jiLHCZ/UnSFjNVnilwd2ZTcAq+11VicaRfhkNZEJ9AVHS/h5Jp+v9OPZKQbTtaicOcpiXSy1hRo5k7+ZJn1VQYNCppY+r6Z4/4SUW8IPz5LicznpWsndtM7fnxN5/UV2ZNDLR5MxRLWlW4Y54d217D+FZZhzmKAT9cQ1AXNWkUK7YmWJU62lHeKs9E3yDKoTKmf/Xr8hKH4HYTq7bab6RVIpLaeSH0J3MxkeJidgjpPmq/r5r9K24W/YtmKmq5Os5dlGlo1kmbY+zAEO8DxXqKgfkSPU8f1e/A3EMP99QJpXo30P2plIercFvtNcmshbd+GbnXa2l2YgQ0TN2an9esMls4g0t9eKmWb6lGYCO/HdAZyvgLUd73TbcxBZSpor7MX+BzDDyn7/rY+485ZMerPfqmYjdiLedBynF+GmvwvMYBMzdULOZthsX7GMdS/2p+al2cymkxFdj3PtQhzT74OYcxYZzz743mtjmc2X3afXi/AkO/pblSH7aTF8k43Ofg15PWJ2DTxlq9yGv4Mc5jn1Rg5VG47MMKMj9vr4byj346p18+w2HWeNTZFTYZzNugufiAzwAI6wAeHR7VXuRmro4fPqU1Z9A85/Pd6XT1n+vJXPvg/zwgMKnnrwXarbimleRzrrQLg6MJNM57bAfltCsR5xyFag3QVrP85RzDpkQ0PbHw+jhTk3iKc9Ntz0nTb+nGGsDesMmhGa/hKcqmN+ux9nxSojGM+w9v2YbWctZaVEi51NF3xVTH2Cqpu57BA1G5TTBoqofZD/kur0eYDWRoRxhE58h/FeJ0If5nKKjd+G9QJsNcQI/Vj9sQbn3INz6HRIENVfcEpX7z1ZUjcJq15YocUVqqod7H2WUgJl/BBadcI/KD7hiWWq3DVpk5/lUxxtZ1nMiKfOmkyDFusz9Rct2N7RyWqfVMin0oQHK9mf6ZyN5YU/tb5Z/sagOOdVtf4ErtjutjxRz9ZwWgAUTlFjNV3Lwx17kLvwvU6+nH6/zjkU59OLKYVR7nTwOXn+T/g6K3hNsZSmVKeuh89FihQb5HKdSZNutE8a+YxYjHaySmTnL+wati9Raiec6m97xbLVehb3qe/S8qUrxWpF1UiTiD2LuKD4XUjDjoOKyPEor+sTtXqqdyI928y026q/eFPq921UD0D9Pbss9+FEhFLoncmtrw2kWIvpWD/WAd8GiFpnx95vwXcEBXQ61ltHW0FYC50oyYsryLq5hmGjS2+SarUNnhAym6oTq/sa7Z76+G909/Fee4relbZln02pbagJOjnVCr0hsMUq2NrwTjvXfP9dyEtXVJf6Tb4hZa/j1LuIhSyYLsU569HUKJKXH1Sc0ZRxbLEfQUQJlrmiBDLO7zxBZKpIpDNmMr1fTGQzGCZif5Dwv517P0vtH5U7RsWTbsLqfzYQmcsfhFlQi7iZn17n5l5R9yu1lPuXsVgdi6j0l1Equc9D1OSkXHlZ4Vt7V8ioZ02o1cPOPJeASp4cxfzjBJfOMlcosCTjAFni3tni9rJ3vRf1FyfVGGspvTEVDbEYmYwCOO0lGbdl8r3UKMUZM0kbAs8xXAe2S0NEIVl8W1aW7SptUQs9VMQJKaL+nqdaU8N27DOdJjhb/X1CQQPu2cg8r3O+k+GDrgqr68m78OCM6hfD7t7xz8ZiFCFPShS5LVavUtpm3zXJYSo8OfmuCuf+8W6lJYWulYg9msmiHlo9ONy9E6jT5vji/d/9RNTZsP46iKz7KV33Jm0jlmMQltsq167liH3uVgLbDbuwNGFjWZxgyTLJ6gkKMlciaxqnkzVxZj5kTtq9c6E4I136DKfJfmGAyeLpZTlULE1Ow85zSZxnaTjPC1H5wms3o5QjpbwCrzMVWS3sOBEVKLu699HpOeRtOWjaGVJ5JhrbPZLiFqfMzcrfbSiGUGT41DeclZOql+wl7UUWNU+SLi4VMW/hUwu6UeU6k/mn2xpOnO2pek+T0aqo3NVxR5/YOVNq75U7Pz45Fzjridg+UZHnE3vqZI8y9yOsMRFRjyBPCuvQqV0Mwn6RL+mQr6IWafKKVnniZPG5StJfEve9s3DFvRaP6vw1aifFS4tTWDa7gmNZR+ekYtXWmN5aiV8wMnmGSVhnsm5E6ni1xmcUOUra9KW1qCrB2F4NA/USy5LTnRtxIvch59GC9aKAkII/dXe1cyeFt7U1fXtaWK5xW+czuothjEueM+D8XW3Vg1D5XsDrnLf3ahVb9c6YyvT3jDpzbF5jSfmuQp51QSxX10m9U8+G6eWl/PSiqeAt9RsessJHhUw9v0b4D4KfZYW5epak0JLuyFUxLTC9oNZwi7OoS/saqp5Qf6/IXceo4y+9MN+49Dkzzl8ayuMMRQ5PrVp1n5FXLCHVinSZp1Ul82QnCrLKK2YVRciIS2eK0Z1SU7U1mAUsK9sZPare29R7R4rH27UTAUP/NdordZaEqs+n+3sGag2O2PcgTnlgUT7qP05FJ3JHhahNZzJDPZnBfR6v0Fjy9DJxHpYq8ZMlKSNl579VnTadeuelfN+y17lCqn4QfjuT3jIyFFPoTER7vPens3URPnjIftckonpTnnnF4rViR9vUNlfxeagxm4JSthfs9IGKa7QEzbp/EWpikup4Z9ZD9SS86F7iS40LSdrJ29whda6gSKPkW+6YTukIXQohYNXQbKUEFck8E/PHY67VE961rLFRec/rfLYM6mWziOeGlZzUdG3qXdnJ4fwlt2Jfzm0BMJyJfabO/TNixxrjFXGGA4VR1LXEXTOUEU71t05UH1ytABJRfGkvhoi7/lKe7jmGK0bpepEdmZARtF2tARUybGqbxUSLPm7rMpYHE3VI4nTDHLYdRwk5vb0pdEZLMIK9EutRR9BmT5PDyWL4bzp9LHatfgeuI6vsYfKyn4ja5FauCfF/5exf2d5Ew+/7wzcNv/u1HfC9ilhkFmtqf07+v1lkb1JD9iGfIvuS/chsUgt9ziF1pB6GnkvmkQPIgeQgMp8cTBrIAnIIWQhLvQgmshimTw+PWEYOBRQcRlYAMlYBClaTI8iR5ChyNCwpLUVo5WUQNCnSSY4lx4FoWYsph15yPCa9BsggWUfWkxNAmH+anEhOIieTU8ip5DRkyggP0A2jaDqdbATUjaKQSwMxEx5MycOSjIPYnAChcCY5i5xNziHnkvPI+eQCciG5iFxMPkMuIZ8ll5LLyBZyObmCXEmuIleTa8jnyLXkOnI9+Tz5ArmB3EhuIjeTL5JbyK3Q923kdhtPdziwdie5C/7eTe4h95Iv4Z378O+Xyf3kK+Sr5AHyIPkaeYg8TL5OHiHfIN8kj5LHyOPQ4lvk2+Q75AnyXfKkdql2OfkeeYo8TZ4hz5LnyPPk++QH5IfkR+QF8iJ5ibxMtsIb28gr5FXyGtlOXic/Jj8hPyVvkDfJz8jPyVvkbVJRHoI27YBtH/HDrO+C8R+CER/XVmjnaJ/TPi6rLVtR9oOyn5W9XXdx3d/1Kn1vfbZep8/T5+tL9GX64Xqb/tW58+YOzD1l3l4HPl+o+Phj6E+Hmd8Dc/gGwP1tbRX086+yfaGf75e9UfbzuvPq/gb9zNL31ffXdexnqb6K99M/90Tsh3z88cd/Z8j6+Hf8cx/693+/Tci/P2Z3fvnhu7exq3cvfvda+HvRu+Pv7vHOZnrn7Y/forOndN4GdLMOPk8gJ2nwqf1C+wD+/pH+0/6i/Yu21nZoO8uwpzJ7kbR/aTvx83/hz1dhneg630H2gLW+FVb5TlitmwCf18K6PwX4vAvWbibZjewOa/oArMRzMH9KXbcBff0QKOx5WJuX+NpoQG/bcH2+BnS3F2DrNVylTSQAPHgjUOEE0OHZQInnAJWcC3R4HlLixUCLlBLLgBYvA2rcApR4OdDFPUCNVyI9VpJqoJH/0VaTf2qHE0s7gvxLO5rs1JrJv7UWUtCOIR9rbeRD8jetXFujES2slWkdWoXWqWlau+bXjtN82rFapbZWC2hdWrXWo1Vp3druWr+2m3a8NkPr1WZqfVoQqOkv5B/aHtqgtqe2TttLW6/trW3QZmknaDXap0GqVGj7aCdq+2qnaJ/STtZOIn8mf9f214a02dppWq1mUNrW6jVTm6sNa7oW0+ZpI9qB2unaAVqcUqLWoKW0g7Wkdog2qi3RxrRFWlZbqJ2hLdbyWkjLaI1aTluqbdKWa2cCha3QzgYqO5cEyQyyQzuSfKQdpdVpUe0gbaO2QEsD5b8N1P5j8nPtUG2z1qSNa8u0Ce0w7Szkq6eRw54BXv4e+YL2Y+117UrtGu0K7SrtaqDc8ymfaRdpl6D0pHT4TZBbHv/TSEUZATTu15E3MtG4kew3ktnWVDK6qM8czieMDH04f8ce5KMm7aOlZR8tLd+xb0VhfiHxryv/dZ9v//2qd99/dt+eD+0f+8Fe2r7QE6zlDJCp80BGrgY52A0y7lRUBmfB6l8F8uZewOKjsNrfAdh/CFT1MqzKz8gvyR/I38gO8r+wcrMA5/MAl43aSu1oWN8uWKcIYPQM7SztMzCra7WbtNu1r2iPad/VntW2adu1N7V3tD9o/wB+DZTNLPtU2byyhrLGssPLjilbU9ZfdmrZSFmybHPZ+WUXl20pu6HstrJ7y75S9nDZY2XPlL1Y9mrZW2XvlX1Q9teyQrlWXlm+e/l+5Xr5/PJQ+Yry5vJjy3vL15WfVB4rHy3PlW8uv7j8mvKby28rv7/8ofLHyp8sf658a/mb5b8u/0P5X8v/XaFVBCr2qti34sCKRRXLKo6p6KjorTixwqgYrshWTFScX3F5xQ0Vd1TcU/GViscqnqx4ruLVijcq3ql4v+L3FX+pKPgqfNW+Gl+tb67vEN+hvhW+w31H+9p8nb5uX7/vBN+JvlN8hi/qG/YlfZt85/ku913ru9l3p+/Lvq/7vu17xvcj32u+n/t+5fvQ92+/zz/TP9t/oH+Rf7n/GH+3f53/ZH/UP+rP+8/2X+S/wv95/63+e/0P+L/h/67/ef/L/tf9b/t/4//Q/z/+fwcqArsFagJzAgcGFgUOCxwVaA+sDQwETg6YgdFALnB24DOBKwPXB74YuCNwb+ChwPcC2wJvB34b+GulVllVOaty/8oDKpsqj6lcW7mu8tTKkcpNlZdW3lh5f+Wjlc9Wvlr5buUfKz+q0qp2q5pVtW9VfdUXqx7NJ+NLljQvCeczqWzaiJin0hvLlh4WMTKpZKORyDnu0htGJJ8zG9OJ4YwxZjbmI9G4mTGz8SxcjhoR+lY+wtrkI5F4JpIfjSXMTfSLwR4OZUz2Ygr6ipjJHFxn4slh+MjFE1H6aCSfHDYy+dGEkc8p3aUTbTCAkQsnhzuPpSAtXbYiinfM5HD89GbsuZlB0ZwaTiXNjc20Z3y/OYwfrewvAtMq4Wu1oWkVE2zDNmHsNMyu7UZhNkiYDdKBbTpkbx2R1OiowZp22C+tGTIya2SjTvtBJ77fyfrsZH12Ii6Olc2PU/rsQhi7EKgu9T502Y3PuvFZt/KsW8yrB0frUXHcw4busZtkE0Z2BL/1sb/YXZ/SXT/e74+Y0XgiYfRLOPuLGkGHA/j+gPIIl6+peQBwMogADaoADTKABhkuBukiDiJC1mOP6+Vw6wX9rUeC3CCfbMAbn8YXPi0A+bSNdANHNdhABhvIsKnFMPGDkR4jXYWaJe1GRMdRbGNipya7thuZbBCTDTKMbYZlb8MKWobtl0YAMyOyURzfirOe4qynOKIkbr9yumy+UekzgTAmEKiEeh9eTOKzpJFOZXOZVHrETGKzpNIsKaaYQhBS6jqlGDwpu4kknAz7i91llO6yeD/LCScrQc4WNYIOc4CFHPaRKyKeljwClFcByjOA8gxBebqeecTSOPY4LocbF4QzjnQyIZ9M4I3N+MJmAchmG8uxU4dip8bg//h56gj9E6d/Tqd/NtI/Cbwdx1twIwHwLj20dXk0BTPONKYSUSpv6WcWMJSgsg+uJ8wk/dhsskcAP/3IjeO33EjGxO+xVB57iMXH8Hs2vgk/ABH4uhkfHsnRi2ScdZCGNaBCll3mRlL5rJGM4tdEPks/R+NJfpFP5OLpxAS9jsbH4lHswDwjbyToRcLMYrPhjAkSF6FI5keHzEw2PixBh1sUdPigoNMPBB0uEHT6SUGHTwo6/UDQ4YKBDhcIOnxif1EzSfuDD9of/cD+4AL7o5+0P/ik/dEP7A8uWH9wgf3BJ/aXzQ/R/uCD9kc/sD+4wP7oJ+0PPml/9AP7gwvWH1xgf/DJ+0uz/tKsv7ToL837S/P+0qy/tOgvLfpL8/7SlEyWLFvGPpYOU22boH+YxMArRaPiV8kT+F3qVPouo116ZavzhOsNKgPwExmXXiH9sgv8yJjDcUqmZpR+OyNvZnNxsE9T41SXgyhMGKP8y8gElR/0bjJKO4OLUX4xnI8nskDniYQZy6nfM4gDdiNhjqZySgP8LhqkjYyZ5A/xWjwYAkGz0RTv8W/qQ1N5ZIoHbLpwE5rL6ww1NuhjXIIlq74Mf1Yualqq9xr5hN5iZJJmVgfu0VtHMoAUMNf1/sjIOIiozSE9n4yaGZ1iJJ+FFdcT8QglRj2XshtFUnpnMtK4uBUWAdgybiT0gYm02ag3JxI6gpbVQSaZmTHAtvAIdOoS6NQn0IVPcBe19MPg3W0lfwA/ao12pvaU9k7ZfmB5H1F2bNmny9JlV4CV/deyneUHla8EO/qU8mz5feXfLP9u+U/Kf1VuVcytWFtxY8WvKj70Lfed5TvXd4XvC743fO/5PvL9r7/KP8vf5e8HW/Ua/3Vgqb7qfz+we2CvQF1gfuBQsEePAzt0AuzQawM3BO4LPBb4TuAHge1ohf458FGlvzJYObNyr8pPVc6pnFfZUmlUnlN5SeWVlddWPly5rXJ75Z8rrapA1afA+jy46piq46rWVZ1aZVadWXVB1Weqbqn6etW3q56ser7qh1UvV71a9eOqN6t+V/Vh1T+qfdV7VNdWz6teUL2k+sjqcPXa6v7qDdWnV09Un1N9YfUV1bdU31P95erHq5+ofqr6R9WvVb9Z/YvqX1X/sfov1TuD/uCs4Ozg3OCi4KpgONgTXBc8JRgNpoKbgxcHt1irf27taQ3Vfzdg5Q/9n8KhhZUrC7sV4nUnBwpn/+Zga0nd5a/WFGZE/2odVv9CwDrw0b9aPmvGIw2FFXXHBQoLT15YKK+bWWPN2v5HK1T/lYB1aPuvCvsU9mw+pBCqOz1QWLJ1ibVv3eVba5qNH1jl9U8HPnzwtZ+9/uhgobpuXeDgWPioumDN33/07J/fiT24+kv1kRt9p960+dZHZn/3K7c++62HzznvrrqtgRvyG284vvaA4weajviG+UK+/pGzfU9O3DZ+yuxPnzHaMTBy4w0b6zoDm77y2HnP1Aa3PfbdrVfWd/vjn9m8aWPtxs23PvCZ+h7/miu/ueGV2mCNdeDWn1jz63eMr96ZChQCw6ctWTz0mOWr25FesXNzoKBvO9I6COB5+Xs33/lQ/WuBTcMnbVpbW9jnxN9aq+jce7f/8K/PfuOcicfqXg3cuvHUG3tqCw2FGUcUrqofDARrnn/k5lvvrX85cO5YZHygdvmpT1ifqn8k8PcHtr7/9P1nn/2Vum2BL54R/8L62sJuS8KFjvqN8MpbP7zllkfqrZ8WNrRZ/sIpgQvHRi4eri0cGP2DdVT9dwLW4sf+bM199dHNm79W91rgluSpX2yrLbTMm184r94MFK7606HW6rqZDc+d9F7dHS/WDKbve/7Ze+957nP1b1/vKxwaOHtz4txYbRCcqh29vuse+8JLW2Zf4d/yhWOvPaXujzt7fL/2W/pHnb7V/qadZ/guPO3ik7bMvtS/5ZyXP3Nn3UE7Mr4jgWTeWlg9648LqoM7VwTOzg5NnFbbdfbdj9Xv2GvVzr0Dqx4e+E1dsL161huHVs96vwE+w3Purn7q2Wf7fS/4V1iVvg5/8MZ7vvnFx2q/f1tLtfH1lcevBy1Qd4k/+MgN99z7/OwXT3v46NjGiy4Yq1sf+Nytd159b+237j7nxOBN1vqaE+/KffHB2V+549avPXJbfmN806bUOKzJ8tN6Vx7z8AlvPnjvF268rb6gXXraWGdtOnnN1efUrw9ceMN9F3299sNXvmetqP9awAqc/KPFdesqvn1H9LD68cBF54ydm6g9dezup5+67f5v1M8shGu+73/Emr15TtA655nCwiv8L7zhe+Lp+7/14uyfnfr4qqGN559Hobr+1nuu/Wrtt+/MbDjxnOqPflfT5R/oG3zOd5z/nUKlb5s/uK76p1a1r8kffLL6pvpnAp8976zLzq1NnXX9N+s/Klv176rAYY+knqsLtl09sKUOEJt6I/18XWFwx3G+uf6ZNe88/fTbvzz1m4dPXHDZpRfUX/N3n6U/Uxjwn1tI+K666a6r7qr9wT35wXXpZN9o8nPXjdefEDjv6WrryAtrnnj4trvuqL/aH7yh+kGr+ojqoL4z57s4df7xW2Z/1r/l/BcvuaOuaUfOd5h/pjXr9jGr/zXrrb9tGQ9+6Z7RR2offeDBl66uH/TnLzn//DOBFx6+6/rrbwbALzl74pKx2tPOvPHbQIj/3DbXmnF4YUZ3xw8L7wWWP554tW7mlu3WRdut87drL75j3f3r8musl2qeL7x07JzCj473W/MKL9VYF1kLtxcWWuf3Vtd879Hb7rm7/ir/rLM+Z/pe+ShzeOEC/4158/MGsFFHY2F1fS4w64nC2re7P6y7c2vN6ujz1h719wWsWQ89897DD15w7i11hd8Ezhw7fcKoBZkSuvq+F34521oYeOnOe167om7Wjm2BWX9puyJ390uzrVDg3eQLhdDVwPqvXXz/+UfNLhwSWPeFE36+ue6zgdWbz90wSG8cnTiv7WKQTi/7dxz90cW+Tn/w4ed8Qw9/74zXaq96taYvevvXz6i3bt1R77sucVVsaHaDkGeFZdZEzZ+osPub5bus7qmHN8XvqS98cWet7+L7PvvAQ7ODz5Rb759XY4VBcNZbSStd2N2qLXTAf7sX9i+kChlr90Kt1VH/nYrCIYeAuDzBOuH3VrkVshb8qVBeOKFwwiEFX+GQ+uCSOQDbh997cr11S49/ZM6a6ppt/q9Zx/mW+YN3P+hLfeXRs56qvfzlGjP5+bvH66111p2+q8avzKVnHwuSs+2oukKldWjNyyBRt//sgS+dn7utrtBfuM13yS2X3X7X7OBBcwpn/mGBtbxupnWk3+p7z5plDdc/FLBSDR8WOuoKRwABt55X/faO463Gnce/teP44KMXvmqt3mot3Rrk9PmooM+/AX0+K+nzSkaf6xX6PPfWr134SG3wtQtfHRqqDtZ88NTL1n5UGeh9zzU19XUV6upGA4UDn1/367qnK47f9N3fRuYs3nRyH6zMG/7fWJ2++wNL4O9R/t8VNviCCwEp2/wnWEHfUn+w5oVn77rjrvofBMaHTztzoHbxyd+x9rqs3vr6E35rVc9rharjT8qdlKj79rhxZ2dtzynnnhmvD9b81L/IOuF31gbf0f4lhU5fAgQtMvrlyOjP3P+tFzijX3DuWN0Jgetuvec6xuifHqkOLpozU9v6TvnWfd75qHOFP2itPO+1j3YbD9a8++S33nxp9L4Tr6+/PBD8YWLOi2MfadufnGMdsXTOoxXx6rEd618FjB71u3ZrhbV49gMBa/krVuMbH9S1+w9fsaoQKnTOzgUKnW8XDnlrZd3Ml/Zpqg7+vGC8atVt27HHqxeOB62WLVZwwdiOd/4J/Lrj+lX+4I6FgTu//PCNj9Vee+Nll11XH6x/NPDr2x9/5uk7T11Ud2ogCOo/aF2wxSqz7rBmwzt/PfH7y5addGxD3fkw3af9t1913edvqQ0WVhfu8p1iLQ8EtYfet155vzxovbu9cPlH2gr/zKesnzy11x0ngvi2moKFsDU7eNicmRe3V8NLg9XBTSfW79w7OLbj+G3B+AbfQ7ENdx5bWzgSLYP60wLH3HDkV9fWHVMo9/3YH3zH/wtr0Gf1FQatv2wv3GJdaRnbC4Y1bt20fec8WMJV/8687b/ZuqXGWlq4xfd7f2GFdbe10LoYhOfMmpf8z1rVIIoLG7Z9tM948Ld/XOyfuS3wwu33ffv2my797PV13w9cc+6ma+O1R67fcEB9c6ExEMSn37r9ZvF0HJ4etY4+Xdvxo6DvUH/wc3Osm98pf2mfd6yfWjcXfkpXsvBk4WbrSVjQ2+e8bM1JVQc/Js+dRz4ma88jW8b3en9rEFTHU39/Vrv7FWufreXWyqB2/y+sgV+WB7W737TmvFUOJPSq1ep7OVB41WrxHRawzi3MKJxb2MM381t3XpzJZs5de3F90Fpw/nYr8Zr14HZovsdPnrT0+uCFr+yofSVY816gcK410zrX2s33LvRRaPF1BmYOxRLhS+pn3n7GnOCFz1gTzww/Y216dq8vvWbVvGbN2G7t/Vqw5ifWImve64V51sJjCgsLc1ssUB+P3p04oi4dCG6/fU6wZuuOw0b8wVvmzNzxuY45wduB+z43B6B+6Z3gTdWPW9V73f3HQ3bUN3w46wOgzg/Oq34LRDJg3frbK4W/+QtnVlh/exUuZu5YGbjxnkdBWz8fvPXWq66+vu7JwKUXBS/dsm3HzK3BvzS8fPrLewXf9b9k3eILvl1/b+Bnd3/50S1zDs/DNE99pXCqH6nrS78ALL39PpDuhVt31IMU2e63Zu3w+b5VmOizllxKuXD9S4XARyNW4KXCev/Mv93+RHXws5+9sS740ZlHVRcaDj2qOmhFCoM11gHbXv39B+seP7g+CPcXw/2Z2n3vBK0Hzq+xjnwLbdUnA0H/2urgRZ+eE9RufvHPL5Rb4eDbF241x3ac8Epw+0fa+DVzguXPPjQnuM/bxpzgxd+stmrvA7Ts+O13qsufWLyu+hvWgK/RH7x6zu+sOdqzb5cHH/nNI5q1+2NvP1ZuRYPandZE+VPBx+O3b7iyLmh1nf9qdGzHr/7a82rwguesK547+3nr8uf2Cm47+5nMN+qe/ik1dZ7oqLmqUCgs9c+8b4518rvlwdc/AAvuQWA7a8075WBKWTUrArP+WKgJbr/w5eN/bn300+NfDn71o7xVqb1vnVceNAKvXe+bqZGbCM1rXk7I5wh5ipC3CXmekPcIuYXmPskPCHmfkJsJzZLeRWiW9F7MgGp+MkJI1QpyFiEtc8gDhHyNkIc18m1CvkvIs4T8kJAXCfmRRl4m5CVCtmrkNUJe0ch2Qn5CyOsa+alGfk7Imxp5i5BfEPJLQt7RSBrTr4ZGYoTkCMkScj0h9xByPyE3EnI2IRcSchkhVxCaJt9CyFUauZaQMwk5jRCTkNMJ2aRplxDyICEPEfI9Qq4k5DpCUoRcSkickHFCmgmppAlf8itCvk9IhGifJWSjRs7QSF4jm2n+kDxCyFcI+YaPfEcjTxLytEaeI+QaQs6pIBdVaJc+qB1DtDaiNWtamGitpGxlVNtBncNyEiBfACz+nhS0GVqdtlgb1V7UPiirKDsYnMPtZR+Wl5XvWb53+fLyzvJbKk6qeN2X8E34LvFd73vR9zPfb3wFf8Bf50/5r/BvDxwRyAReqGyq7K/cVnVYVVfVKVWjVWdXXV71ZvV11XcFy4MzgwcFQ8Fjg/3BE4InB0eDm4LnBX8T/NeMihlHzLh2xs0zbp/xwIxvzHh/xj93a92te7eTdkvsdtFu1+/23d1e3e3t3f42s3XmupkjM0dnXrv7kt1bdm/bfcPu8d3Hdt+6+692/2CPrj3u2uOpPV7c4/U9Pthz9p6hPfN7nrvn5Xveteeje35vz5177bfX0Xtdvtcde72+159mBWd9atYBs5bM6p21ZdaTsz7au3rvffZu2Hvt3p/ee3jvLXt/Y++/7L2zZkZNXU1zTbrmszW31jxW88uaHfss2efrn9I+9dVPvfypHftm9n1vv6X7nbXfZ/e7Yb/79nt69t6zx2c/N/s3tdW1eu3RtXfWPlb7Qu2fav+yP9l/0f4D+98z58g5x8+Jzdk059I5n59zx5xH5nx/zvtz/lm3vm64bqzutrqH675X93r9WP0l9V+s/2r9D+vfqv+tXqkv0lfqx+mn6GfoF+lf0O/XH9Wf13+t/31uxdxZcxfMXTv303O3zL1p7pfmPj73B3Nfn/v+3L/PK5s3a97seXPnLZ539LzBeal558y7et518949YN8D5h9w/QG/ObDqwNkHLjzwCwf+/iD/QfpBSw9qPyhy0DkH3XbQgwc9fdAbB/1p/rL5LfMH5p8+/4L5z8x/ef4b89+f/4+Dlx98wsFnHHzRwTcd/K2Df3TwBw2VDfMbuhoSDWc2XN5wT8PjDW82/Knh4wV7Lth3wecX3LfgOwueXfDGgp8teOuQwCE1h8w/5KhDbjvk+UM+WFixUF94+MJ1C0cWbl54+cJbF35z4TMLfxmKhm4NvRR6J/S3Rf5FcxctXXTcohMWxRdNLLq6saqxvnFFY3fjzsULFjcvHl984+I3lsxf0rDkuCXGkk1Lrlpy95LvLPlZE2mqbzqq6dSmRNPmpuub7m7a1vSrJmvp/ksXLb1/6RNLty39xbK6ZbFluWUXLLtx2UPLXjl05aHJQ7906DOHvrO8bHnt8tDyI5d3LT95+VXL/3jYPoddcNiTh/11RceKS1Y8vOLnK7WV7SvTKz+38vmVO1cNrbpx1WuHlx3+mdVLV1+3+t0jmo4474ibj/jFkV1Hnn/k60cdcdTJR/306KOOXn/0l49+/OiXjn7r6A+P/t9jgsfse8zYMdcc8+AxTx3z52N2Nu/VfF7zluYHWma0HNRyREtry1DL5S3XttzUck/LV1u+0fJ0yzstv2ttb+1rPbE12ppq3dR6fuuVrfe1fr319dZftv6+9e+t/9sWaNujbW5bqO3otp6209s+23Zv2/Ntb7f9I1wZnhM+OXxv+E/tR7T3tqfaL2z/QvtX259of7H9447DOlo7ejtO6RjtuKDj0o47O57o2LrmnDXn3z/TWn70nJ+Avt1/25Zxq+Fd0Ov3/dLa9N6srX/d8WHN3Sda5CnfrAfv2XhH/Baz8jP+WVuv25y7biO4482NheX1+UCh+SerrT2s2dv+bDXXgU+ytNfyHfvTuh/dn33iSDN/+9gXN92w31z/rJ88mHog/eXRSmt+YWfNGYHC3OHuo47d8HVrXh0YfTMefe7Nh798/rl31H3euOakdbNnfm3Lg3OutDbW/Oh+X+EytL2s0YC18m3fjwOFlat81i9W7KwMrN3om/n/kfbecVEk3cIw49Dd0IOjO+2wK7M946przhkjZhQVsyAqCEZAxAQIiCBizglFlCCSJImIqAgIBiRHRQFBxLDK6q7renqegvWr6kHX3Xvf+9339/4xM91Vp3LVCXXCoDGhbtqN7pKCWukxyFDevUyhxlqhI1NwjXrETLejINQUPWLmrKTkp9l4MNjOQ80QfnaL70+sfAUhb7483BrJzm9+1ZmXNy5gC4CVwKJ6aV/2hOAyl41xEyndceGn7mytNuIE/xyLh6PK4HQFyN0xu+T4XGSXRoE+DczDIugExtOLkLGmE8iH03XNNko4/SvzNG5Nr0FrVwzXyGvRWDewKBF+KIe17pLsWnCql2ajecpaWE37hCR431JB+7IamKCJY2DKuLxu8+3d3Derd8MxUxqmtAxVLlkbFOitgWdM4NmEsxmqiisrkInGnkGc7RIztTyIvQYGfzMu8GkIz31Ee1u2KEex8jN8JagUUW97w2Mx8/FIFmeuaq5Tjmfh3v5sKMjexgujy7maDIcJfHPH5qkHea5pJwsvWqTKwTyE74deVhkw+M/PevnfMoBgjdkI7mAjxGL4kztZl5beE/jIlqVYduRmvstkuHhRSmM4yzU8l2jOQqmuIqMczz8zM91DNrPQhdQQGwptIVwUFzRcZKgoMqgPMFxUp/68hrvUKZ7FICNYb5Q3gY/SVe8YSuoPCm1twLEetxBUb84iE2SkbZQ8r5Q+R0bKSuHH8XSVfqXQbjwtz4aiCkkK/EeaAkVK+E8F+g8tfymcd5c8gN+lWNI0XcoiNznqvr9JeO2uqHg1pIYrqYMlC9lAwQ8zPlz9gZZpSnd+4h72ZlVulTRLOHaBhR7I4gIrL8A7w7UMjMthibskROgpTRNMMSvniYypUZPKbrBNN9iasol0MBhTT8qn0JGwBMvVG9ASqhd98evzCBo0E5VNYqletPwWagOjphUudQtz03YE5XT3cHdF4uPPej/76uV81mvnq3cn6QnnHwrdURtlDs11CT3xnoJ2x3OrgTYB2bzrQ2PVAQcprmN9OOy9FhDnHe7bkVPWFzLup1zOLDtqgDOOFlBzaW5ifcBMCrXdNWMikpggyf2lla7qswGbUMTiQ7ZnVoR1nMZwHUMjt6Z65ewygKGwQDnJyc7cVx0OB6mcTSHOU0yQLbxSlsZfLjyidkfu1MwLW2Pum8ih7jofWQF8xRvyxWUC66Aso4Wu2gBqIo1+azZUvr5Z/qcmiLFlEX3L8pX6wF3l0k0xrZzyNcIpr1umFto291a+ShcBuTpblstsBb6DgWMJMFenA+cyR25xXoYF90F+JSvdtO2LFFHPvOtA9TTyIZcCAX7KvMzw5BQN55a0iHpLm4Jhwv1zVQfrOprRey39F3ktRx1hwQ9cXM3NjOsvTJ7Puz1wkPn8LrMuLotbquZSRtgsmdEH76yKcdBJva9EaWp761l+1LWsBA0MR56U07oVnotVXNxcwVeZcyE6LSViw8Lpa9dbrVH7B1KbErbGpJnI0Wqyb3S7Rt6FYK3UZ1KwR0ZLefEN+tRLjyCjJexM7TR3yWc9f7zAUpirnaa0pA/1HZfusqjWztQu2qoqfXcf6j69503V4ugM02u1V10yxy6mdJsbpj+UwnQYuo9H4/vDfEk6tLnCy5E+7CyTZMJoaSbsVMLoMjSalqPR+5skD15JHxi788NI+wchRnoQHxksgCzXvsFrJIfevmVCaank6gu4+FwKYb5K6PYY9GGD5jYD3kPeIDXqNLQL2qq2YlDAi0HAqw+UKbuueQ3DNJUMDLzc0PTrleHIVG3GoOFrh3ZTy3tjbG2JR2r8VArGxk+1lutoNHQWZcOA3ilKnrK/XMiukET+CiMapZHGgMrRmUM0Rgzdqaq32VXA4F09qxIppttsc1+tNmeCYlNOZ6ieJ9oOH2AzZdZIDeqLbCj5Zb8KOFsO28oVuKea8IYpz7mPMA4LQcrqh2AMetbpo4M1hxhOKA69mvaLCeiZ5yFqzhJfT0dcZWB0/KlbqoeproNH262YqOE+LpmYRddfXTGo1/yVg5YsO3POUTOFcYu75l2mkr/AR358mbDAXXKqDixrpZBgXAvjfy3rKSzA+zu+eRk1Ey4wyLJMeYgBC0inDqNMWp6Ehd2r5bC4HK/wCF+9ys964b56jVKxh3zlC+gI3SbnI+NNHnv3eGtOgy0FW8txOTf0iToaefnoFVVTmkPPbsuse1jZBwY7aqYyHrG3thWq5B99yyC9ApaVS1IbwQuv1vH7yqGrnkJXTTwTFJ4SkqWqSlras/eqeZM0k/RBUfoWVKCcWY2YWTZuGxw14HrkefQrVUqK385QTSFzZovdicUq1AUZjUQxmrkMMi9BMuiilh8JJTc4imAQhtdx9eXGDTRXAo5afaoECRMhYjcjhzZ+ZUJJqSKxzrIe1tRNrecaYT8enOrR049NC9PHXdFwVffjMnNqTX43L0TtULup43oPvb4wd42aa5ywes6cvhgFPZwMP5K9NHpxWmnZ1fR70Zq3y3LWjjWZsGjNBDVXdRGeKYddsb2cb5J7ObG4PH655SyXlQtWqvEkVMDSCkgvk1x+Dl54Vi3wlh34Zw3EaO4wMGcatEGdkfHkAagT6pA/Eph7N0MvxqmR054JjkhP5bDs1HE3zTTGO/S6b7rqLrxXjltx/df3SZkPa5PGod6atYzXZgePhSo5GrK/CeyaJAWvMP0ndy9gp+/OoyH4xMHSJknhqytYnMUpYaAvDJYk460hgP4uvgjmlr4qkWTAB2kGfDzVmg2Xa6W3QN+Pd8Zk51KZ4vWroVX5z7mqd7BEWXWLTg8+ez1NXZgfHl9tAvTEjM4LbXdsc8Jb9WRs1Il0VVHq5vFztm1wWKGxnO26aZIJ9zoAncI0SuiKeuWCUa5lIZTkgzpXkVb821OhbTFXJ6hQL2VpZUh4Gj7F2xymbjFVoel9foclsAMoMAU7vKmRPhqFei/rO9HJL+zKUU0mlFET6IAN1G763FbnMytViBr3E5qp4TJ7p8yuvpkYFRehPkxzdatORG/MVQF1sPY3cDCBBT2gC1qDdqAeaD4KQ4eBQwysV3OZjnBAae8eVfJHBDDPLkX6+ARqVqJDlMfWdW5rVPLRvOKU4MNV9Wa51z1Z+Z/7y+BoCXiUKwrqYOkzC3ycC4xhKg29C6pABlKrtNGnNIfxea4Kjc6pwufZKh/1UCMzuq6ljRKOPme4j43XnQcMWWPbXTMYHirB4zEGrk3ZOHSI9bqRGvlBVpKEd8oNv2k8DAthMVsX8QIv63QeVyBRCtdnszPYo/wyXp6Cz3teOfiXH+UJ15cHNuXIBvzhaDl6Qcv9+fM8ZPt15zNZL16+nrfi5Y47K7QMxnbL66TQYec6FgwW4JoEE+OtbDxmcAuMHXm4vv/NeDfh/QtFQZVgUs15FBhXCcPowt1guF99iN5/AemfWq4Gw+Z51DMaZII1dSY79M1+E5y1G8mmmzQPr9ZOQx0YRDcvpbg0rxk+Y3QaqSqvfDUyEJZSSEnL7fgDLLeEVG1Dc87DWC54OG/FruUhyU/ZiX+GxxVdBpHleOgrGkWONvoXGhRVD+FH6DwlH/EaNJGuQ++UEFnLvExY83Pf1XYDNHKYpZWXrOWFy8hAGf4Yzcmgfk+4/RuMN0lnYMvgT2jwF5WoLYNs84Z8WqV+fI+oRD/AcE0+Az+n/Q5SNcj2pfRAw02mM6jX0h5EKXqBP8cWGPdnT7OjWXluaOuFsbnxWtaa3872Y+VguICVnBI2YdqJ0Q9Ya6dRI2n0fbMXtW3hDgdx+DuubstQo+8FL2owLXdiFWe1lpwr9DfuyXNJo3gwwhWECCukIcb19G9gTYVfDU0V5/ScY4S1+ldMRV7QaAo+tGPgBTWMHoqmUm627vaiusjtyuY09QCYSuFTPxm9oojY0CozYBZcEtEENu+kEcaY5+6FjAq0awokBdXSAmQUzg8mRDYYoqWwEVPZMhrZ6ajsFzjoXC2NQkbB7H5hKfiOqlbEQwkXGi8sXchydo0wgOJCn6MBVCmN2oMvNZWWd0JtciR7IV66FzOAEJ+D4ml5KtsAw6RgLtRZ8JasIqIeRtZyaaY859GZ5dJ68PJavxKBLyECyG0sgOCd6Yxxc7dbeIbHZA2Ya79163oscRw0bbGnfeJKfe6pDpcoV7ocPeqpAYE5cT7hbKqq+vJKLOMsZ1DXOSsmqOUBAXw0LwUf/248mvh4AnBqeTyP61fE1oNTDZcOa/zN+Jn8OFaODPfwMKRYUlUrrTKuFea8Ke5TROeFxLBrVMMXLe4s3kQzcgvWheU+dIrkud+yetKOPmcjzh4PPnlSszPkwp4I1ZvKbMIJXtke5RWqLs1H3WjuRacjHlsObVT1mjSri8abmRlld8lNzf0W6rjB3WWVycrwLdG+6t0MrtJmq8NqS5OxJdbQQb23QGm2Kio99WLMpUDNH3RuiKvt3DVre2u4X8wOCTLlolUxBZpQJi/uUk5WostUtRszc63zfLV8C0tm7ggrvW08jYXTprTci0Xz3Xhgn0uya6SnkfUYNkWbPo+NPcJDAJljKfyo3biZ14mpOTXSHOOTPOSHLuH93RV4BcxqwKy2w49czW3jo+zZqPigK6rMc+5LNc3KGmEEExyTeCZBlX12iy1OqBU0zIuliePn22xet1y9m050dwpeqLKy2bRxrWY3zaVHsVtYXOkRlqu6jad3BN2P59Ki2AP8RLZV9CzCK69r1qmOa7rdKnpuC03adkMFxvkvYDRmFmDclMLO85dt3rQRt8HVJI9hYGLzYqXDqtOBXhquCRqZoPNJwddV1XHWqK1mBcPVIMXKucPV8mqxjWJBXSQRl14KdVFKtL5GO5oJXjs/aIIKfT94PLLUODJo2rMFL29fDTkbrjYVxiovR+8MOK9BvzPeHo6etqox9inQQZPCQKfMa1VqeSCLO8ydxkOiuVx7njtdyXC5DiwXfXsMI18RwGMRvr8/Ft/7BiSwMC9gLG/JKzJrYSbe9vtYzmMW3vZz+YT9v4+0ZzFd6YTxwGjjOhr6CyepUTSX1B9VUz7W/mv3m+ym9+9M8ElX94dqT16+klecFl5zSWBifJbnSn5m5WFHeMlpob30tPEzuhA4KvzKmaj9Jgfo/YFu4UvVBagDwRo/wRn4Ho5gUW4OMnI2pzAy8ju/6YZ6BrQlGEONjlHyGLfWOwUslkui3oLxO2mU8XjWHMkLoa4Y6gsVp2sDq7GUnoLkZiz6ieychlJpA8Yapdo3k2k5hgxlkXw8vxz2lawpUcTchVF3OdcY2KO8S4cePhUUouKSImPWpqmuXY4rP6qZQ2/ZvX2bp4pzdV4fb6uyW7F+7E4NprOlDQrQ1BTVcE8xrX3Xk02Cvq0qBZyZfeXlFQWMu1Z7javD2e97skdBrRzOnotyY7e6rvfaulEjf4Gcb1VcSy63ddPKyxVJje9vnSvg3mFa+jPPPUXWJd3BaKN6IMO9TyPqnk1E3WPNgP7R5Pi7Jty7tLWhiw+przD7jh7df1Qln0E0L0d5LHy4DeXRopYm5UAWk66K4pXFmDqGnWV3VHGLzGCE1ns1zy32HcPjtxUst8qsR4u3kpA2TNhOmRHKdqUBk/RUPLBrmLCXXuSBaanBhM6aDPhB3RmcEVzNfejHcn/FaXd9GbElWwh8N3atEHWCb332EqIsxGdVNxZJhKiTrCXb2ICR6paayhqbWu46xqtbMV693oPnUjOFjCgMXBPER9SvuA29a0fd5iJ9McQqXwwS6YthLvsiIyFVeZvmLvqeYblYs0ksfrrF4ozduGgOmfGI+qN4yv1quXRS1puUTSdlb+qdFM7h6Q+OdmM9Nrh4eeLpDxCnP6XEgShwFNF12bfOk+knqBhPf+cMs+eb1f3w7O/grZnzeIWv8VeZfUeOkMmeSRR66TxMNmUDWlw7sbqEcewW6GjBk2dVOt8k1A1m77a4iLkvGnAD/njcrvXcf+C0AGNZ7pOoZuK0oG5ZiIFS+Num7OSWNcra5Zhb4nMwEjppSp+ptOCJxuhHojHaJtQpC+i00huTqXzaDH6kJtMBLavFAhhhsg2KnY+9a70fY4mtnQDKxw1jaGQ8RPzlXue1zBIBCZYDSRB/o3bdWdahhrtnduMLuos0SxnDcKVmmY1kku+a4UnO88WTfNcMT3Kp2W4+jAULU9a1xfUnXl4uKtzwqCeyLkKEBX+AjHo9+4GMOqnFBUMc0I16Px51AEaYEEpGrRvzB7ynFoogeMGj6zzPstvquALfoSzX4Is6CldI+3lkkSv1JpFO3CIZeJHJhaI/CzNN2XEtrlNZzhvjp/S5X9L3tSb05kOFVxZiqsqf/Uh6tL3F5VtwMleZtZtxz9bXcqlw/u+eXS9tWYght2LI63P5m8K4UrDCTOiUR1KYMkX5SNh5kaXlS4Q56TD4luTKPdDck8IYYY7yUWZOReWi7LHjFs2ZND5r/iNNur69y6WbNyNjr16NcrZZ7Oq8XCPfuaNwFd5sYDLJ/aI7nhqQf9Yb46vHHYL+SF95jz79gTpC378SfD5Sc5/xE026/GnuVLpgErkrfEfIjo5FjNcJ9xMbjhhwkSfKqJk0N8t/OoWYvUMmoykmTgzqnGnWsEU9HwKUVxgu8reYguoj6qXN7Shu1vLwLVmvTOS5+yvAPx+OVkpA8kx6DQqUBaPpfp7L569WzxTYChR5kElJDE5+YVIOA/LRAJAzv9+ILkuK2el7UV3EBLqvPL1A1WvwQtRDgxTTGXkCMi7VTitVhD1xeAI/P13yhKuCechYCW2K7oBelnfyqosaruJOeGRqnslL27hRseoxMdTyYK/Q6yY5sRcLMy77+ESoi5kzm1cRuzZqrkV/y5CVMRs13OvFmNuYYDL8ukP5evVDFyrbLcjTymT2yvXTLZ3OnFulnspwVZvjrvnmqrB8bCx0VdpucLHfrnY/ZnGBsjnnFnHd5PqF6LRT6oeuVJZH+CYrE/kYsAdPSTJ4EJMYe2XBrVsFhVbpFhaLF0+bnm6djxlyA22HLAnszivMk8JuTC3ytAcKWw7kaW+j1TTqNH9sD2T0ePYvmqu0/PWOkhVu2mHFmLSWP06uXl+N+bQ7O5TcjZcPbr0ocEqyTtDYxlDOIdvC4kw472KQZQBt0mh5o1vP/laIVzdn1ArXmIYrVwoyL21coh6BDBd2M5l5cU3WKjV3I9mVStp4xs3RxMFpo7WD6/FT69TTGC7dOyp6Z7pK7gml0dAj/lZ8fpTkfHzZ1ZTU/HgpzIBS5dviKmBBZl7Rq9fU8YhFbYvHN2mi9Uda53347VZeTfWtGd26Ws8YpbEsV159dJ6+NI66Rq98Sh1kzh47FRRjkux2brl6Bn11uPLWWeb80QAPzwD/rXhSkn1vgn4ehOWD3k1JfPanh1AwiT8J+sqr0YGBIXib7vDcvH2FytI1qUETw5SHpd5NuuDvG6bOY054rTu1SjVk6mL0o2YTg76/N++lWg5jhZ7VMDhd0pADVDqMzZHCCaGnMh2cDtJ3kRN1i74LTtRB5JQDKaPouWgFtYeeDyswr4HGoBRlDpYf05HN4jLCqsNowquPNsWoU4AGyS/4NEvhoABRfDSrAOEp93Gf8Saea7rJp/EK0RKFC47hubjrWJrbyHPBE3l5ExbeeuU9yofX+IwvrZfCUuN66AWqPKR6BKp8JKNhKLzGQswYNIFaO8ZziSjEeN5Y80Q9BsZT/XA2ek0h1gJU45BqBqho+TbYnAu/F0qCa+Dnx1Jc2WZl6qVzYSGaw/RxJyofXtINNqn9Fyz33eaknsscu3DpRIqqLMzNQmPPDPChhvosXm5qMihjWa36XK5y6aaIjPQLFzKPakqOU6gTs93b1ddRJY+Gzfnwe54kogaWYA55OW4CDAufw494BaDLzCfICNHmQ/BGc2FQp/sjwEh9MF85fXXWr5qbTENMXkneZZuu6sXMYKdZ5mp5LBZznsLoWjx8br3g7T+f59wdWW790oSczQ9VXOIcNosnqMqzjgsGD38lKCp0Rrdc3A2eW+KOxWFzXr4OGZcJl5oUEc9zXo54zP0Gy5DxZBZ1F1ybFOWvhtZwoeX4RL2mtwt+1BSay9r+QenOr0fTSoSdFeBUIrkAYdILQgdlOT23uQ812c9xq4PKw+PwwR0aS8b/TNjORFV5ZgrondXcFYZQQFd0pc8CQ1WeunQxQ0VsdoI1ecwRn81HHVXD5yxDlJdmK1JQw2n5yRyBlyQAxt6KnAU87EfTFvDyQJKYCI+liTlz+DxkNYeXg794nQv6r0D6WgrndfdStr2Ur5v60c29e+HeIjfM1uWDjFyv9OI3Qffj7GgdewXfnWW5R5jLe9WTRd9BP8xgVWP+6tG6zTF2Ku6V/WrnKbs1cis2lQAnNV48y0Y3crXQR2TF3mLIWsyIPYAfWnZhNkvouL8cDpWDezmxbzmLRcwHxsL4f5s4mdFo/hikj7Zo1jJo25tBMFl9sFKJ5DPuwlzNNQYm3azCYlpDM0ZWIPgXmZfC6AIYXEQWcmsdWeoMcPdXVmZevhejmRVze/ULVd7tC2ERGs7ni9XbgKU3oO1eDSRn0DBmbhlqN32Rywxn9V0Xq/hxKqSY0Av1wbIIl4EGFBJj64NFSjN7nbH1bwklVQ+vzkeGxNh69eSxasgSNMqS6KWI0qClNDAHr1zLVck3wpFCSQxkSmPgiBIyC8n1LqwSE4VYMVGILWyOxfz6HjYW9khjhWPh7EdkEc6SlFScknqYh/bIYiovR8sxzAcdiBw57GGv/SG9dpiXZ4Pp26yKW00SGP+KqK0CiFISE3/Je4wsurMwUpsexTcuYAkrz9X41O+q4V73ZbmP1cKmL4wtmuagDZCgboKfFHVzmMDLf/crh8n5T/NhWrnk8QtpiGCsvDtijkUuXpHzTXPBSHUnLSgsXlPKeDqu3DpXNXhhGvzgowHLfDT3KczNR7Np6Bl499ITVVbCdq8rGC7YZeXpeSpk+JM1snXXzKLlqJtu147S7drW90QYLW5Y+W3gSsFTx5dMeySNA04JPZDVI7CCnqWoJ3heZJEn6oHrGQvRbyUPXkgfQLTyxduBOGX5OLCQpP4hTR2HZ82cvFyFvVKwwK9ooDiV2l7X+ahK+LHiDfm6Liqf0rNgHLxWvvpW4fRavai5PT6vQhdRIwWoZdVXRZNZli1mtLJ0qiYuRw/VGv9LWYVzm9sKvZVLN8cE8+uWqbWL0IevGimzrFaVVHqWqJPS1bB0c2wwTzRU5BQa6E4hoqH7In4G0d1i0aeddktfHnVu+YWIPqP5AmB1UE7QYy47g73VoCio20IEtv8s42HVLZEFF9eYVCDDFbhot8zjR+pEp/WkvBV/NG0uu540SAw092i3KHPpm3cz5lD36XFgQJnTO1t+2Y0B8AY69HhbnRfhvVVaLx3v/cMX3ju/pWy3WIvMij997ThrRSyp1/JwgMhqHi2f8LG3Igp2nGRKxpDQ8kdrkgwnBZJeNYlQ1qRT21nw1xWErrzcmnQOJw3QFfzYmiTDSQdJwV9FKDQfi9/ZtdLsJF4eJ9SBSlIH+6V1Qt1JNk54F8Q/xBjs9EPhHWG8R2C+2wqz3SMw1316Nx8naEskn8BW+knQHsJvDfmSX2CrFFYIDcoupDZ4LoGRT6UwUoB5uLbKq7xQ+S9RK46IWq8VMJuIWPUwFjkrud+wTFV/jU+3xjLWbzt4TNBfwhkpeAh1y1hcYMkdVzcFsKWcHTijJUrRQlDDhaawyfFxcfGa2TRnt87Dxn6uyYJElwfq1kHV40FB/UhWHunMS6D3Oyn0HsTLI4Ua6CUBc0zf/YSalThXeA8aCQz8QwpnDrGElVgK7TDxF+pussRATib5Ezphhl+75Th+92qQvMa7RgoBF/FwtT7FONdO+qfWZz/up9aNGD2Mh5+lMF7rNpe9zTdqtyzib/PweAiB3vxltjYQ4G2NEhiFSfYo7bYQkut8ldc6/+tWgLSA6QRZEdeHIolIwiTCdQUpvwU39gsclYKn9de2F0JbKSzUum3BVGzYOB4moAnl0E5n1XuA5WpaL66bqndl6i7FpyaMUYNT80bKLmDDFhuV47qjx9w105jtEbE7L6s+pD14r9EuGNs8nxnOQ8+dFcsWs4qiOvCp49KgcCdmNJJrP2iEwOHNp5iJW+3NHZ0PH9+knsL4RV4KuKriPNC+Y7vFO72vF3ryeL7kzlzo2nqjmQAOeONA+77/NtEXDfRFU36iDXEhpvxCQxbNpd5ZyHAJDSMZLnUIas9ENdf/zyb8j5F1MfTS3RA+hfk1WPK8gqyVL+JyazTCQqbBLtn0i18IV5MyqvnkU4ZrCuRDifHxVGa+22ZbL+9DB30wc+Z3NiQgViUX67NnySVlTu3HmkR+36/Kp89H0NDtrw6t95Wj+bx8Ha45At3IqSCcgQILGna9+J4wAG8U4JD1HRj61T9CAdpSrklYkswGbN/nq3J1Ox59OfFyUqwGLdZOxowlWtwyjXpCZ8E8Clzo22getWGjk72NyaLLzrlq7iOy1UYp9wcuOLFWXYh2UmgyXQj+1MmEU7f3m7TePkVi9AdjyAVUrS+5gfLtybpC41B+EypRpicm3tLg9FurExYvXr3aWo2zrRLXpquv6G/0OhdFssiGJMlkU6rl0aJdL2GqPwm2BPk0B2I0ibnrdzf5gwSbtWqBvjoALNH6zmUPEqRGcoQuQ3i3lu3T+K8pMIzgq5stf5C0q42E12Fq/1HDQu1ufEb+T5WLuz2EIH6dykkwHMKjzS1blP/QPKXmE90TTEola/KFTLRgyDUYciBB/JmNx3hyzWxLTiRStzzUJWcVY1rhB3acsIznPlZm4RUsIAvYqRwOl2He7BjfUCoVBhkTvuwwLC9r5cvQu5ZpsJtx3LRi1UKyVA9O89v3+alct+DVTUqOvaiR+/Nx+ed5SNnAnmnx6c7783g+8PsYMhvpLX90x2fOmY9rtejGGf6D+Iw4Aqebo+9qizGmPUw6m9FSj6FnYgnmUTno5c6CuYqSOgiq45qChP3K2837iSV6EHpUJ+zPzp1PZ/IoiMnkFzCgRmOUoIez9Bh5Jouphxf/67VF/Hq+vNGKP5AZQvRveflW/FVx1tbzj0qt+OU5ShhVikZhFmIUngdtpbsisNS7jqsPRLTSKsHpwflzh4+cEi24yXbe4HECb6KS5LikxBi1dsrwFkuGq9+w2WWVjQmuUGSat9xOqIu/zdW0Oj0MysCI7h1GdDUY0elUf4Toito/jkzOiJbf1rFg6MznLxavOQfxMD5VSXSBxPS6gNx9PiIbcxrel448ThWsYTQGjBXClFEwmhpL70CjKdS/OQxnJzjzqQk8WAzikXGqDhz2v4VYYpb2M9lrUK5sfDuIRlubB+NsO0J2MS5dBkfF29XduA0RqboMY7nzw3E+7ivJF/rjvRWE99a3uTABt/bV0QKDJYAbbhj3XoRKxVBFw79pI5DgcuTxrzas2Lh8onPcwLb46OuYA8JCFFqzlS1/itxCajHeIdlgdwcPYBqhTh9aCsQMHbNxhFS7V2Qj0nic1CoSn8cicQIWiZdhkfj8RB7aCgvmshd4fNzPsSAedbQAd6U/TsPHB6dp8RBX4CH2Zz/gxUgnYCvxcAak4pSRvJ/Wby57ufViWygsV8Q+vYqJ/tlnXAveeJjsf8Zkv+Uan8xAx+NZ9aBvAvKZiUMvq63DKMuL7hG5JqVXEivv3/DyjlGXM0HrV56drELcpCmI6XR72UNHza0tVIFLrJuZyXAH+yHz558LtVNPZGK2pnrd2WUwguE+Y0ai1VbxNdk1x1s+/sTKLzvzlypwZ7bj1R3ER13CSSP5fVo/jER0KlI8sgweXg3hzfDA1rI2vA2PM6A5F5aUZ+CFaVYCg5rroBkMcpEBLAHT8ubBE2EkWoIYGvNbeflEt6rdrt9Fx30RHi3Xmi1t+SSyY5gWPKiLBLvoau5jP5Zr/lm7l5CEESKTA+bF2vbuigdPy0tBVs35QEuAsijl7lUwMRHcRjVvYJDBKruB/ZdfA1otbBzRvJVB6oIx0GmPOpB22eK42kbkgHQeFFxGCotluJT4hCsX1LjhJHIn/qDuGqYC16u5Otz2Z+1K/X/yG9a68/igbt9Zdn819xgDVSJj7W58Ht/g81iJz6MTMe0U1cAeUC9uCW+8JXryXNoo3olgkNa8gyIDPgUz4F/yiGB1lLDc3sLPF/meLWUkJ11XStZa6hVhU5H+1xqxPCYefVErPEFYt0iXImtNCRbWHWeP87CAdARTwk244ChWjlOsyQ7HKfa6lCCM3CURTUW1t3QqZRihHYeRmyVbkt+NPSXEEuxmyb4qUUTU7wTbdbXcDVOe8+zMcjd68FzKYOH2IT6C4Pwf8te6CQtLMdinsu21XA2GSj++hnLcbL9ygcmiZOfc07xvKwnXcE8uJyQlxKgFGfN20c0us223e61RWzInLiaeSFM9ubh5osaa+cmL6rPNeskAE67mv/rRck90nrRcjc6XVq5TbFwjXjX1R/A67qjlMsxwF3zMcE8zzHBXb/me+LduYw1MnMvGEjXMN5c7vnDflAVJiyu54vEldzy+Xy55fOewSSyYm7JbW1z/VkRf5qFpJLu5+dW3SW/JdfqtFpe/0wIC+BfPo/mCp/9QY88TPs8jlX6TtCyZj+fzoNcX3bZwjHDjKKC5kai4S/LHsYfhR7IqrRoU+HUk6938SqdAwS0UPs0cxK6CgfPYmfwrorE5BrY76rn/9MISe28YfIhPFZer1aMwmq8qy8Isf2oyu/PL+iTGxsdEaMLORQYFnTb4ctMxcOlNkO3RQHImDeNml6C2yGC8KRqrdmXQiNKh0FV9knbe7LTGRlxr9XitRlkatRTpi3cZ9OGkaw9Ucgu8p1xY7o/2xJeoK3sK3pJhWBDdF1HHHxF1QTg77K/uXVmcXgsaMR1iDpHkBS1JYvKL54qTTxdUT3iY+TSl2r6W+6NHJM99cP1f6fFfRv93evwPa/6lx/+jx/9Kj/8m6X/Q468R9OaxW1gdV59VC/NquBqhQT+ruWEaCwZghIWNO8gIb+EGU8xaD1nI6BT/5fhQf6P7v5+5iD/Cv8tT+N0/Wns4l/v4VmhUliTdLDygmUU77t22dZ3KZdu5uD0aS3pakk2Jmnvt0Nz4RddlVKI4C8OQvJb77aygVcKwEjQMC9pXWsaKEIRF+ckdc+yK/WXeGAbR+lZJTvd1B1XHnLxOviQe1GOmzcFYAN2waR1eYe51tnbtPBY30OZdr6LVhYqN2Rdrw7K5lxuF98psmisKO3j8VJCKK78Q4nxFlXwpvvCIZj69aa//jm0qrmiDe8wKFfdy1VqXybs0XFlN83uxMzobB6LCwyLRNqEughU1eAX0OFGDZ/NXHywkRWFaimWMVqMH00G8vfZcq91DBWaAc2qkedqoofzMv3jlSZwsSrQ4EXzEI9S2uQYnozmii+hGd8mdGum1QfyBVJKoxoSs6Bk0EPuJq9dY1K65K04masBfMHo4giVEsDJlkSPRPOr0gOlR7NvQJbwvWCBzd0WFuMBx6w7zDS3TlLU0VyCMYDBUsM5i4hdQ6Sr5z0gWrWuu+6aSKtwdUQ7D+U0wHTMJnVK/ZNd8Kd3aheVC3WoWWf6jEwcwTpjI7hWiyGEKY+EUuRooFLWPB3QI4DTYnqn7HhSC9hxG/L+3TMFZFd48WJHadouQWUQz23qBff6fF9gJN3humTvmrsz5IoE6wQeyLzHo7VouGFR4Rvq2rFnEkjSVLi0HD3BYc90itjceV5xoV1FwZRB/MJVMSqM9zxVUMlyjA8tViLYVV1kg1hUaU3Zei2tnXr5CZIpSiwQVpjiPgzBTtL2G+/M94Yk+YZ7oT6Jefnv4TvZbEzEshZpr+b8PTKHm/hRjU2Bx/oL3pR2XdxtgSvIJ80St7hq3CB4KbnEh3XHmE4ibKO6iwSA+MqEz/7XLxCDEJ5nvG1AE3nfeFEse1UiD4LoyF02niXe+6MSv/qcT/7g/7/z05htH/j6zGDkuL3QoLktfjDH+oxquCs6Sverd3PB/DAPAcK+/iQTAVeliAfR15pMSWDAaxN9LGsv/882Sb8r/VvF7U3gjkvV/K4SfCK+WsTiZII7M2gCwxYT8Bs72xNk3emNyv1ArOcSX6eiHvT1LbGAUpb613BPhZ+Gk0rR5OP3fh6Pgau7ciLgQT6IPpB+3pxw3rWyV/IhAJLq0fpXvCXOQeDkhWo37UUcYs8zak5igH6nlMvVg67/vzLh0X5sWiiitfckQfEW1tSSIz6xdfZZ1qOUizfYR6w2cF2mGB3DZbITwSbSNIGr7WKK2v0jU9pfNdvMryWyI9jhpEI439gp8ys7yXNXPrJxkqVqz5pDNgbq0rPo7k+jJwaamkhiiffqqJJ/eMouA1LSW/1K1aOqDy8HFSgtedP5R6Qx19mkjTv7DG+hXbYQFf4yFQabEj2YZ7s54Fr+OJKcWJ0wQE87wvz5XRL1NeBr4jhOijOFRMXI/RHPNeKqU1B1UaA6Oexjk1sxRnADfCzfnsfCLX8nKDfysEugznQWnei5RWK8zz+NCwHx01sA5l9nZKiSdtgBhLnn9fDYsNILnEls0jM+lMp97WSwGI2Z6icRML0lnpsclEkO9iWo5ZhF46FMyazrGVlXGM/i3xb2L6LzQGHa1argVMb9D/TCx68MycqHiSze6EnDoX0MMxqKytDa4Mz/jzgRm6cwFL7OWq1jOKWse6teHbZ4zg/+luG8GgyF3sGKfSn3u4z5haF2nrn61HcQgrb0SSvbwM0v/bmheLVfsK1w3rqExUxfIh7jytZtY7pHvv5oo9t2B5Y/be/AUr20tyr0io4L+JC7DSeRLc8UePPcIj4d7JTjBSGXtc1MaF0GLSqD7hm/L/eErHEaLWu/X8pAvLuCLi7711TWIa9zBLBBv1Q6Kt2q4gO5iLUV7Q1erEPJlakl9T0n3tSvpFTyZ0XffdvtpdXR0SWjonj1B6qbmbjrX/teXWMj+R3kyDgJP42nsj2vaxrZWxMihUZwusj3mY0goJBOV0Xr516EGT9XT+WxoWAQvLyRzswEvdpNobKmLV3ESbae5mq08XmjYDiPnsYVfp4OAfoILeCJeXtJNxHbc5la+eY6utG4OvL1a5+CTX/B53RTgbXslkoWj2q1KO1bXuNhrJnDdouOzVTCo2VU30FeX2MJIVlAbf81D0ulkJ+NC4pS09McDvBZAQixs42GzDd6ovkKgC89V6w0hzz1Y/DSAb56RSLzTfHey84m7mRxu7syBnzKhfc42XhhYMqeU+1U4dYnn3vXn7Ta6L9c4L6K5X0ewzWPgMi73507WTlfsNW5pU05qVlqmewgurIApuCyMwc3+eUS4j5v9024IecbN/mqHm3VHCRP45s7Nlotx3UeIY9qnI/EEIJ5Ov+hus8DRY5HL+hMn3TULGVzAJyRmx3WVHOpwK1uyr2Y9yHqfo0jLfl4KQ3ELnzsJGbiFv7KGkGfcwl9ZuIX5MbP4taiAdO4PcUyJ2XE5sOm2AszLBUMyshahCI/sr29G1oJHlgKXF/LvUDMpmCnOHzjeUUTBVjDGbf0inMdNvcQt/YIb+oTb2ZgYxqMfW4YR+Ku+NyEwB7g7YHhTkpCNT2JTKdDZUnjhq4Qp0P4foVqmoPatoVraIxVM+TZUy1uQQu9/hWrpCgplCXRMpyuvpVTXOiSN1CCGLkEzlMDFM1fCfa1s3D3tNfLwntB/esEytzA34SUoFZeefNbr6KuXE3bns56xr14td8gXfuhJXNz6+Z5opMDw6IMS4gykNz9teIyam+l7AVyTAy5tC/XrWMRsPbUh0P7IscIf5tI7Z1LIYO+MqZ1NutyzK9moPrNrA9qz7LBD0OrzHc0ZTu170fPytlu7DLgffdcJfsrcmIS8I+otyI5aEOYZkWOCmcc85XQXB0s/dTi4Uje3nFs/30R+z68YBhZVFsHQEklFnRTjW2XW+IXmd2hkeq7WEmSq7NSg81GaQsbLea3XAtWg5emgt10DE4rQ5EqYXIQm0qA8detSlSo9Yfu2RAwXvH5F4BIVovtaI3MPzTxaHry/BNYXwdZiyS810jhIUeZOoQd7rlzkqLYEoQSdP8ikxgcn1JvkwU9F6Kc/mDepMUWJsTt3XFDfYwLdV522UfUxtUZdNF0wB5O8H+aZFxUXKcCshksCM4ItXPGaKoqQohh/d2GgE5RSo+hJaBy1YYbbPNF8fmv2+nz1BBjnyUNnVEp1MYfvpqLv8Dct/8O3SNhcJImrh+OYy9+Dt4i8sQ7sNZcZWGL6DCmR0rQvslavZpDt0+FYPjtYpBy48jH8qLnPgMnl8ucNVyehTmpLBvErJg5Wy7uGukmAwzVxGEe9daTRwGmiN9gZSn5if4mwskQS3gQ96qThxvCsBPkeohtfUW+ablb8YgKSuYVIamnr5Y7nhTkTc/n0TVVpwmozs7ULTQdocEd+puShqE0xhJXAziLFH9A5pHZGHWbPhqA2Su4jdKgRvcGsbk08S7zBXheEJN1sNAHKMhdRVot9tzqqZzCBUXGnM1XlyY6DTZeumkK8waZk0U/THAZ3n2s/zMbu7FlHjQUu6h6X6lWgkr+oKIFs6CkJegqzq6WQ3F1ZDeNfl/QTFk0m7mC21CziDja7hLiDzYBM6jDKpuUhqE0JXC+GZWWSjHIYUScV+wd85Sv4AX7GXCS3yW3vnm2aQHCgwLsMZdHu6E/q2MXkI1dUv6Y69PzZdmGPRStOBa/FPfGIvbkNi/KlvsVwqwTWFEvin4L7UynswKvUrvA9fA/GlhWImmOzeaOzBlwPP73yUnX5qp9/qCaPOe22/KSVCpkgeqjO4WviXSSDbuoDxcrhKyuhJ17gs+HJIVmqh0nLu/dePXuKRn5ICClVhEHTqGou60F35TOaCwUXLUMVoqbJELybEV0NKooV8U/mPIVFNdOfco3gjRkIk8LG928X3piQSBzB4m/m1pr8Oe0BRhfSKeN7Drm25M4qNdc4eeUii0EmqM3DKdBOXaI/ccmVh2Vpt27Gav60SVoz3sTCbul4NVcVDnXKodecE/NN8hOT7pdG282atXH5wpVqeRmS4gkogfRiSZRuDiYgqRJ6Nb2CC6In2MyPqDP6zqI7Mkbf358IhlkZ4RFRarR695TFXVUOS0+ecNPMFP3AslSZ8EE5ZtWV35sS7pQ+TZ6Ksdpy4gfmvhjj9oa9PJe5nOXqVuFfD/ybxnKZzqxcq9rLXyUBaTBjjLMn6pF8PQ/yRCD0nFkuK0s7H/1H+fxGyXvNCWYV3+nGnEZ1tr7dlqhHmgtMZXhkSsqF9RPUGxkzt/V2avkzCKuQgPVDKViDwT5efM2B/0hzIOyLJ/MvwIBWkkdcmfWBWcrWg1Gr6cjsVwQugH0CE4kDWg58wO8vTvHpdcJgIq0LTN0u3Quk10pz6/z421jktimB70okYQIvLTDOpyPAhvqF3oBsqDz6In5+XDKFPgvfUU9KJtMXYBo1g96MplF9W5834eexU4ppT/QdNWZKCV1RoiTAv4gJcgeIuA2VWbtYiPBkxdBWrx34fmJoKzThcxvHvz3MG4o4uwbRuTx0JyvvBbHZUHBbAlPKpRcgdiEv7w7hhEwSR3Ku4hsX8pSvHuRphE5XYTJdMYKVTxLaZsA59zgeYqfw3SEiGwZiqiy5iqmyFAwhYhYvnwwRWTA8c4M79MpRiHGNsOQwAiKU9VfvPdVgznlV+rhV6329PdSgf+jmufuqyMiA3ac1mcxxn/VHV6r6TFiEhmnWkHBtelcnNan3ZSkXu1+rxgVDmOOBYadiValhGxdg2Wihm7ODWt5Z2/5OKu5NWGvDi9yhuy681AcYCBFTMPeGyfYHTLabJvMYIudjJu6cJKwYjEswudXOVg61mzNi3GXrxwmRp4JCNKhNa4S/Y2KEvx2tEf6yYBiJ8GewNLe/etGXCH/+Pm6+JMLfhWxdhD+H3dlQgRdltyebjdckXVyTZeKa4N3wS5WkoUjaYCxGTurqQGXzvfafYo0H891FRgXP/2toi1mT375hTV6TKf8SOUBBIg5xe2CycSHNmRPf/ehW132rNTyXZs523+XH7hrBT97lw+8awHbGbaKcVN54Izs5wIkP6IMzvFlh1EDeAX13Gypu72LRd7N5NAEZCcPGVpGdEko8JIpozg53kdF1ERmdYnEi7iXqcAc6Zem6iTq0BiaLsTmhwXvktyunL0bdMSnQRVf0I3EMj3+Nrmi70QOPZiEejY7NWyiyecdFNm97SIwf5vEmuZEBxvHICO8q9J0f7hkeB/rOh0ff4ZGAPx4H7gUeCVI48UjRh2R6s5CDBwOlusFAKRmMtneVg/t9cd+7NsDDgzweBncEDwQenmLhIRkIFP8931C8+L+ZccG0dZM/JJu81A/XjbsDpT48lOLuaMfg7uC6cHegxImHkj4k05vVbhvIz93D3oS90puHefjuq11a2h/StMO83JZPxNvwCmbEhQtTsADRhZWj/v7F2mEkloUU+viv4235pnwM9A4DPRWB3nVhdwtnyYWGLZGmSd6f4g3OX4N38n/D4FxoAxqSDVcOsZYt6f/OZZ8rYuoqn96t4d7jyuvErPdd2KVC0jy2tRfkllHXkWH8YG3iXDFdCIVeX1OfCI0rSaed+YQv3a4dxN9NWMcjE/9iOCbeMd6uTb3Grm/ugxN76kb3oFZ6VKCU45icBAptEkNUxoYcPxGkzmb2bN/2JX4lhDK1WdQjxnQhBbGm6CkWZV75F6/Y8I1gl6jt6J/BknujPdu/xr0U/tSJd+7fyonrd7DN97DojtvKUcs/6OoZgiW0dKGfvzIGt66rhbTu8qUW4j7HebuyYhXVUdEl5HaAVJMuRtXUmUJH5YBUZPXjs0Fd8qYM2k7ikUTk03texrx5T6qcNoMh1GEm8eTFuNsmjx3iR6oRjXNGK0GaxCSf97V12Oq1SiNf7ycazEZ+MZi18yP2sqFuwutWi9mScuOv9rL1or1sb8ghcQhmSDMhRwkzytAMWr4T7hdKkiFTmgz3v1pReoiJQqyY2GpFibo9/2pU+Fw0Knz+xajw+RxefrIev8VDlRQS6sVghta/gBU1ju6PzCknZjqahhP7gfVrMXGAmCg/qasBl1nwHGeToIqXmAFgjiFeI2sKHUJWrbEWdcm6WIvyfgO0byQ3nklh1YClfF/yAgPrpScHLGF7D/jq8jkgnG99Ex07BwSzFv2++Hed1fl3XetnhhND2X7j+cHztBcl5+CM9MU8JczRXqQm0Gh2y0VKPohknCV2ZTHzlGi2mANzSA76kWS9KCVFSrUXJ+Npy/LLfp8zPxdC7oBBjuJy/tMqqMjn6uClnzIvPzgkWXOL2bbGcuskFRowqh7Gg20T/AQToOf9vqgr4laPnLE+ICTqsOY6xFBT6V1u1G76jPeGQCdVz2l9UX8Nlzn4sk1RStyF6FD1IZqrW3MiwuOW6n1lA0yD0UNAjmYge/Q9GoG2IRfQ7wmzcYmNsE7puDX8zvOQdxWRob5+JzROyIny8tri7aKSozy57ynh5HEwOxl7jEbWR5lStlZWdOigkVHtwSNGbWFBB6GDMonTCyeRyaboBerdlhhLrraRtOnQZnEbnzZn2oS0SWvzUNpGaiAdI90rPS+9IL0izZUWSV9L/9L/Xr+z/lB9S/1F+m763vo79Uv03+h/0AdKRs2krCkHagPlQ+2iAqkwKoZKoq5ReVQz9ZmmaAVtTJvQ/enh9Gx6Ab2M3kNfpnPpSrqW/ouRiuGsezH9mOHMeGYWs4HxZIKZdOYBU8I8ZGqYJuazgaGB3EBpMNhgjsEagwCD4wYXDKIMYg2SDdIM7huUGDw2aDAAQ4mhviFr+L0hb/izYR/DwYbjDCcbWhjONrQzdDBca+hs6G7obehvuM/wkOFxwwuGsYbJhmmGDwwLDR8Z1hs2Gn5kJSzFdmC7sIPYcaw5O4ddzC5nHVkPNoA9wJ5hz7ERbBKbzuayFewjto59wb5lP8kkMkZZlPltXNCBS2+A0f+kN3QV7L9RFTIHr6TmknjQPe82Ym5JtHrQ/L9YPbglZHqUq2THV1NFQlv6/1/dLNvpREIQuK0Jm6NC+uPHo4kaJwZZNCx4nZN0PiJMfXgh3X2xy8D5C4JObsbi29aYnG1FqnCIVpotuwrGJIyrIjntSUqSt9d5NSpgtno6eiwj4/kaOZH7YPe/U1Ue+W9VlbP/raps/79TVYb+D6pK2dm8kxmic/qpRUEz1QXonC4SwElqDD0T5VC+s3foHNP9k70fqCdBNmVGT0bZlM8cPztd+tXt99QzcPowWtbqD55OPAuJY2EPXiZtRNbK6oQbr6M0MF4ookwxbnxC+TntdBKvAnaF+aSqe8JjahANo5qPe9lQMzw2uy9XeXscPLhdM4fxD7oQEKqS2W12WrvIZGq6/TPC8dq5hqUkXwyPOqu5uzHK285k5Xr35WrZta7/tvfbSez9svabyGC3fydWdsr4CZ1JzAXXi+aCz2hko51GAk7mMg9Co29EBO3Ze4rELvXZcmKNasxC666aqSR2KSwyrodgGmYI43Hv+yM3ymmM9yqdn0uKc6W6L7iR3s9qHk/JIMq3Ky+rg5tNJAjzTeoljfpAPnWEKY64dusPE9BY3ENd1GgUXYdqlXCyinl1aXW/wWschmnkRACorVp1C3P/27d5aP7m/ne1cv/HdNz/cOLQILk6GfP+mTre/2/Of9N8wvZr5PD7fzG7o2X10l6sLIK9V5E7ActYQ0BFTaRliaExl26a5NpdHTV62ZLZiy6uS9yiloGH/0+87J/msDbNM6k32mljm+fTMmIc+cUwkhhFaohRpOy/3lLqrihlQt1C9jXwJCq3dpVyBm050bKcsqCfIxXmBWSCcTm6eoiGDCGTyms2nAiRexiU3pxFybbzMNP3J1bWemFcMLI5l1if2nl5HTq4vdX6NAafr5qGkbRMEl0HSwdh4EJm0tF1px6ZQC/m9vZbo4LVh5jq4MCMLJJQFXOy7Ki6kEF9drnMGGaC+jDTN2+YtFfNCVMZ7nXZ3vCN001Qb2Zo9AzoswszNDqNyn/Vp8hd+RpBwzTI1rNDWVnUBccUVWpCYiEJ5L1xt58vCeSdlXbizCVNEePntH6HjcrcNrzybtDFK8ka7vX9e5EXykzujrFmFnt6LV2mnm6xae0oE6KCr7D6Q+0ofFC+bBpAy5W1Wdcq7qy7uPSCRqb8okKT9cpQHmJ+hQHUQTSYlu9l7ZJHzF3stmG1eresWFFVC/1qZKloQQEMe1SZB8vy52fPz5MV7G+1yZSJt+KrWNlxe6p0OS+L8M9fs4SXVTHQrnwOmq2xYtDkOWaIwyLZPKCUTs5rPO1UVi6Rd7KjIu4d0zw5SS06bhu0Wo3l1d8x3rbpTND2b4eTM++r5EN5uWjzK593+G8GURa6GK9ICS0TteL2PBdcyXAFDiwXRxTAJISrUFoq025RFtIZwGIMgkYjFr9kii8yN+00d1m+VQGw+fMKZcpSuvppLfSE6SbhDJibQt9RderJ9OO+JWg4GmyCT8XAKajvmL7kfw++A2NQIOP0dKSA76E9KG1uonaIQ3LgrBZDe6TEz99TJM7xPbwxcbuTEI/bJS8kvLTEOrf/AOupndVejKxVhyDDg76pbLzeep2C9G7MbVQLKpTzJU3mJpx0lx0M8D3oq1pi5+K0UbOblokOTjr3JpkyCnphxm8H6kXJVHZ4Xgrou8B78rLNK+1tNAMRQ8nqtJIvhnO0HFOpCJFKRWIqNZfp4kP18VlkM9ikV47Ns4obMckJ+KC+3w/cT3iOzdzX26mbf8Rcb+sFT4V4weP69YIHUyGjlMr3GiF8eHM4Y75l7Qx7x2PHN6qnMj7RsTtvqOStdjUwqdq+VhbNg2edVCaqZmTazVi+kolhg2kqDW2dR8IGy8n2J5vPWtx8sD+Wl4k6jkklc0plVaF4W0UGnbsYcVa22Ym66GR/xlql+/cJjSMz/dSCCw7quaNIvH7xWqp5kJuyjIZm7WBqEp3d/Epp6yYOQtZEo03wFzhBHTWARuvQS+onBLtZ5eusIl0kc01rJHOeRDL/mUQylx3hdXboxAp9HpvzBwkh/fVCQ8as4GVrWLnkQR2crpbKRKwkvw2DMn/L/nJNIxN1aGERvKw1zkO6rKs9Zc+UnaDwLGorkJaGH9BG5au3/WmYpO/OSyLKoC8uOV6mPMEKx8BCIjvK/v03CzKY5j8VLzfYFSM7TCMLqiCySvbCjxh/yrSeytn0MqShntMyTGxcWdltYk0iIzYeMhJVkf87qqJsHi8XTYHlg1k5Camou6iS9Q6NY2WXfYn9piz88InAcyrZ/RBcEfS9jVtdqH2DFk4HumUNlrLkx3hYSuKN6x+9En/XRLa/XLheLiuE+mKoK1SA8ZPTNdxvUbJyb16mfE4HgQVljg9vEX0SLPBGVebjsyF6UFmNQ1aox2ToIfsScgVjHqkshZzqTxjdtK3IIQ56U4tlMZkXzu/ecxrT1WPbNp6QQYp/vkUpTCiAXqXmRTISdfwi9MaUGmMzN2H2o53u4e4yct2gdl5Ey4h5BDGukGHhHJxqxJDRJBqMzGHrVhe863zpwHVWx2WYw9B5v2Qwe2WwGi3fycsyiV3B3h2qDZuPxcgC+EIZ8X/hJbJqKUyFImX1s1G0jNyCXtHdgsqeeD9Zk6su/p0aT8toF+RKFdLR4ErJVrNyN7hZAifLZeu2xNlhzE5kcxlM1pmVZ7jLxBA260gIG5kYb1+u+59WVm+npJ9em0lTZszXkzsv3+yi96OevviPm5/Fv2Vl9DrrUeMXT+6k12fi3Pmd9IbNnmHRSc9sruXMTnrT58+1IH/frIMl/52pe2qjZ9j6JMX165709WStT5SeUesTrddWT+7guslVb6H4vUT8the/14jfLuL3ZvHb02nlRhc9X/F7l/h9QPw+Jn6fFr/Pi98Rrf9B+7/9lvyvvkl/5a1vCnFu2uh16OCjJ1m+B4+JwiOV2CvEubsm8ZCES5IkNySv28jb9GkzqM2kNtPbeLTJaPNRqpD+IB0rnSHdJU2RftDvoD9XP0n/HfUDpaYGUNOpA9QxKgWLYFV0L3oIPZ22ovcxPbCYtZN5wggGYw2cDfYZHDGINIgTBal7Bo0GHwwEg78MOxh2Nlxh6GjoaxhuGG2Yalhi+JIdwM7CYtAxNpgNZ6PZVLaR/UtmJDOW2clWyTbLPGV7ZKdkwbJwWY2sQfabEW30o9EgI3OjZUbORp5Ge4xOGSUZ3TC6Y1RkVGX00qi5LdvWqu2qts5tN7b1aFsk7yr3lJfI37UzaternXk7t3YR7bLb1bd71665fZv2Bu3l7Tu079zevP3C9kvau7T3bb+r/ZH2qe1z2zd8Z/Tdsu92fpes6KrYqAhWRCqyFGWK3xR/cfqcnOvAmXBduX7cEG46N5tbyNlxjpw3t48L5hK4FK6Mq+Jeck3cB07g/uqg30HeoUMHkw4DOizuYIdnXvn5k54J/lh+fqXniNfC6TOSuOtREg/82Yo/nvjzEedJ9aZ+LtKb/rlBb8bncrx+a/Xa6Tnhj0yvzecHep6fa/W8cL43/mz7XKW3W6+dZOTnIsko/JmNn+fhzwJcuyGup4NeR/zE489g/Jn2+Xc9C1ya/I9uEG7rE85n9AJwz3bh2kk9s/VYyRycM0+vg1iHRK/bZ4R3S1fc6+74vRd+64OfJXq98ZtMrwt+74r72etzk5g+Dfdn7ecmyWk8sjOfGyQh+DkUP4fhvO900Ljcz/jTDbfaHZfqiXN64X70FmtGetNwugXuxenPnyRncU+C8e85/BuCP7gmfBa7YohuuIbueCS98G9v0i7+tcCjwzMlOYNLB38OlpzD7epKfcKjNMA9W4pnwxn/Xv9chUfXUeKFf73xOz41eAyGej+Ko0F6P/9/7V0JmFTFtT7ndvcwNDjNDqPsgzCyyaKAbOISlMUlRBQRcQFxI6MfqJ9Rn889hkSDG2rEXVFBfajgMqCgzmdijPP8xOhoMsbMZ2xjWmIrdnz2e+n3n1N1b9/b0z3TA8woee/WV6furfXUqXNOnapb3RclBgte2qs03YKaV2eSvB1xIWieCGLSvB1hGZUjXcpJ3wxWKfQprZTaH/erEW5F7sfh16MvG6Qk4jugVIr62NZMiZRtLaWtrUfOp23uGPVAr6WdPuCPQcBnX9wPQbwplUSperoV7a+SsVV846ghiRqS0HiCcx/cDbKYmRJJ5ExLW6inD+oxo5nQuqUvlR5WCeROILdgltL+CGamL2nqCSrE0UY3cFo3O8pR3yjXm/7hfn/kG4X2xqDFsYID4m4VPBCuRtxWYP04xmwDxnAjvNA4ZvudsmMjOCWVUvvDj/HGJ4HSSUtnl3JppVrKN6qgmtbgH1lDY1bKCK/HkTeJvEnE1EOfDgLcF7GDldPrLK+mqJPNm7J541kuVkmLq4TfiT79CnJwF0IZ13vhjUyk0ebD6GEpuKpceWEG/Cz4s+HP1bFJQ0pFb/8U/j5Q5liVTiPhIciupiAmKrIO/HpbeTcUblDJgESitrjmvk5LdKM7gTV0AKgeUaqbesulXv5IWkUPHFu3tH8tMISGIJHBh5F+rLbv11Md8RwBDknlgjGic7wepazO6SujbOoXWdFvkveG74enAcg1ECVH4H4k/AF4ngIv9IBeo/nwJ6Pepch/Pu6XIW05/Dr4J+CfhH8BaTUIX4MXLvsaYQr+H/DfQPo6wc9En2bBz4Y/Cn4ujeFLwWuXKb9F+Qtg+CXqZ+kndacS9O0S+EvhL4MXnXs1+n2NUjRmdWc5rcjUQLNsC+hio0v7oo0oL8nU8zmqcWpRazl4SnqeHat61X1m5BOoPREYLyMd9b5aRUPHeIFLT9C9HUql3RK+nIZbOhtdJW2qBEbRJsYFoye8dSfSRG+Y8RUZr7U8EUEbLk/EWzgHNTe7LNL5qRRtLoV/AW1cCn+ZSsZCGQOeg74zpFK4bSBygwNkxFCz4bSYcpnhrpjlZtHS0tdfGd62UmZgMZLUCTosqTOV0RdboL+SkG3ROinMTHHMMRHINGYoYLs6cwPmmjjfo3KdhFwnRJdoW0ZDIw/y6yyG+kqtPsMYKP3rMd6QRZSCzrb4Dm6EUwm1s9I0G74adR4IPw73EdW4qrNRj6uBulhtHrczkqufkqr7Z2DUwWl2no5jpo3amTaq/bgX96YvaVB6kM6FSavZUiiRUi0mqTG0E7c6PoFel9s2pO560CgJGiVQ9zrQKIG6E3ZOltJJjK9QxEGNCbEVAiMV7H/L+K6En0WvqjNbeBMw2QxL5CXgvEU5Q+alSrFlUKsDDNPAMAV6DUJ9RssnrZZ3aVUPHVQLeqVEiwLXOpSqt/q83tgZiHPplQ/7fHH5chhYphIxX7l9lkqGxEI6oNNmQX+VQ3+VQ3+VQ3+Jxp7lk5rG7eTn+NaKbaqPTT2VoK8RNwb91FitMTsLNX0XQg0xlBQ8YkXNWF0hU9DFGOu94fvAS6qMt8lRq5Y6xhqS2UAbUGIj4p5F+BzCavDcgfDj4Btb7Wb28MPWmkn+r9fa9KzXeqkt00i7c95sOlXkoAJyUIGnCn0y8qQzLubTaB4oufoil1C3r6yEULdYJcfAi+11vs6t5a5txDvwHCxj5uuwzg136fxgLNwwy0rzLl3BRY1+lBkAmlfmTpHN9mqh3wIOEOtYyruzxr3QsPdZOz6say6jrdM6a5ToHL9VZ2DT3mMI11PEWyml7OpAZ1bUnLaYJWQ9itxJWbFgRG7BaJg1U73awndq+ykWW9VodymR4Ee1VAOvRdw69MasNWSNZdZ1G7U3sqbYilx32XnfUCKJ0oJhUkqDG/ponxPeekraNBZF3LdCSGnu3nZ9FVF7x1iLcd/aKabWofRnrVIkrqtnWQcrvXQVvU17co9dE69B+ChaeUzXWzFQLcpPI7TrLmhGd3VqV0i6tjLWQtrimczWaEZSe7nGjsNauw4zazBj7aR9o+yOpTseJqfDq7W+FGZws25Kaksmtl75JexygLeKaoe89XZOrufVSsFaWffrvkPI4xpZ6TJtAg4ObTL4gRvX69rPjU0qB8ldXMfTpEQAN9rxTrjjbdME1gPz55H6ovah2t6FNU6s2mqUkJiQF2PSn8OdxFRjTIIl8AR7RfB7UXv4rM1paktYuyYFu8bNE6zbb0+wrgDde7E2ZU8pSuf6chjbeF/w0WB4WSNXIgesPcAKWUHr3kba7tdEXDvTcpeRZLOP4tg1H2hyTJnsNB5b+qNv6RxoCxJeyKQzsiKLqOUKmHX0Pb4yGzWISB8y9ZmGTFLxJ4QJPG/LQA+gb3XoUX0mjtgkoOlXA+62IDWCnHWIr29jzKuh5QzuUaG80jmqdqHgX5tZr+OQ0rxpwVPibdl0Zo3KL6kMoBzS2/LCKhkuDYtA8bexBvM4tLZyk8bF1OYjD/cIxiIm8doj6Usy26/WvrBiNDQWLOsx6nbMlf5xPL8ucoXYiKRhDIRLkgZLPG9ESky4Bc8JlZG6tsLc1wfQXbBXLkgLl2s0uFjlWHIIP6eUsqaEpNR5PTX9qWlTnLeBJ3Ljkh5+6E+LamsjjZRZTR7fBrSD8DQoCAklT0ITGp8tmwzkF35PtC62/gvaPib8rfc51BJdotyTKrKudFvqFuXVZKEW0aealox/G2tF02bK1dnN0c7qfHOfzJdXx+p7PgMXvtoO7yzF89Pc0ND1XmzBsbGatZW5pynuyMXVuypaEaEWXHsqT+Zenp3SfM5k/l7rTLuHUsPFu/V1TJDTC7WWOxZNS2DbzEyefRozVmORpRI5NoMbHzd+T7uC+qjVucVnLcKWbamVHVhP2FlB1ittMJfC8spKVVxdgAsKzVHArUHDIrVR61yB2TEesFhz12gxX8z3YU7a2TVktPksbXjF7J5A4/j8V/7c38UVLUBJjc0jdd8Z3QM8XvzsX1fYJvtuZXbnrzac/X06UHbmiizVxDzZlqtpr82C69JGOQtZ73vw1YarucAarSU0L5jWFvsAMauly1tQZnrroFL8JZZWpjbv7lxKd9TXqSVQ1Ni7e4ttd3k7uL7ZP6vNgE/CjfPxx/RCvNJmejzWJF/rfnrR0lZo3m3lSylaX1BDx3SHsT6owzPrzFoqx56Pao+/P5bMTl2t/6bMXccUzFCA5jYM0lw0VUTfLyVVbxWyL3fL5aeMvIMrOmdefZJJ/P+OaMtaanoODa75msoHyreBhvTjujt253zzwR7GNd+RfR4vxAWN9yyaqDHeNrOpmeENnxSmU149ko9bPH7z3sjvMVcbcksi217LaF6wxrZ6o5s0lq0+mFMJEf87OJ1dUsETCXje4p4pabuTCrmXzobpQlrO5knl7DUm2mjl0+SVs99S7GquvpBO/1eY/VtdQv0WYPEraPc0TOE9roj/FNLuvjLb4OrMORbl9hpIXsLDyrdv1GiGiiK1xoenrAzz6fw61CmrWql5d5/3yqGLb/Xpt2gacvPa1IgLm55VyZ702b2XObOgZ+dIdaBd8WclUHoT6EdK1w5mHZjI5s5be9odMT0zFt0tHBTVs3xBKzdBRctYMRKYnQl8Le221baeVWyQk1q6dqzPbHPjPZrLiTRff/TdoT235b6dydTmp7rMBeByWStt852z2hXdKbsSr8NZyVJq12Sq5Zycrb/ed55uW2Bs5FRdnVmLWlpGjKQ3wrtWJbQO4RY9f5edq3f6At2iWZ62cWYmMidFGwrMKmbszXhEs7tKBVrx4n0jYnc1dkVis5aoe4ol/3vMRrO/Smhw7ROUGC9n2sVxt9q8MR+/RbwzuekADhEfzF6RFu2fRyxnmzNAKbv/utMcoxJmzojWKw+6JydfB2fWgq+36NlPOYe2BjHrQLM6kVO415G2LrMFqSIn8czritmWvM1EQZ86lZyYaVF3Yutcyd+ZHRnf/Nho3ax099nt2XjvztX5LV5H7DrPKB3MvmdST+b6NVZCz7PCAockRvSpnrz1pc5LtSrhKZXUpJ7Lzb+CjcppTTuHplWmk1mqmfiW0j1Tbeq13CIc4uruBrVnahS3GrRYq/yR1JPFSeWamsxG8EhErZ6Uy0MFMCerZ2Pa825KgxqzlpFTybRzZ8BjRnr8ku9bA6e906/JIEXNOspLTSllC9HcPd+r+iBIX9vmTs6pASs37p7rb7ZUkpraRW2MnZWM3bXu0BlUuMZ3aiJwFrrBRvopNd93n+d0S04Lvr3eJnkipjNE3Eic9i9iR6nQ5Z2EaiI1eB3SRPvF1ZD38voVy+WovHZ/XK3XLJ9FvZWM/7cJ+fgxO0MZX5hXXdpEcsrsyhXBHNP43VzKp+nzWRYtajXALc3YKWqZ1fnsPG/GyJlVavX8eX02T6AOo8Hz2iJ520wU4PNGGsdng+XYQtZqqvd+zRF38+tas8HrEWZencNzMDerDfnFULHag5p4bxo435YM4hsYD1056Txi39G6mBqdTeZ3LGlbZwzzS/75Q1e++ShWIHdhmuc7i+fOA7v6m5JZnnRV0Bi4xunmMhI/hiq9lLkKm3uHaHTRZNWmMa+2Clvf3BaeY+ubvW2831LUTGTWxUXoidy5eTda6ZXySyxbX6XSIpqtHa2NUFvF1alRO5+LlVNpfyvYXP2CbYX+rwl5Y+qeCqtsoYYM7psE96ICT66l4duzHRPIWdGs1eGe5vDr82y4k6uM5k6z7vrZYM9+KCZv2p5RDVp36TwriFj213IF6mrR+568uqWpna/cd09pa4M2U7fuPNQ21kx2h9GsqIvV54V33v06OxXEN3dE9ZeE8ktVe97X6v+0tUyz1neSyu0KMV+LqZbs7jfxlqaJvuemmVnT7Y9vvGN5+IXcPjW1K0jeuedsGy3k4HiQCs3t0hRdb9qdY611nOcsdON3nGrzBN4pFcZarK0mTtC0ENcsBoHZogUn3ovDxNuJirdoj6aYmnPeQdvfozdXKlnAJsmb08ib+ZX77tlFh624zvcUl2d5u0Bm5dvg7Y0mfRSu8LBZZ0u5Mp9nDOwv3M19/r0k24J9s1EjusVY37oLXK8aJ/A+RH6D7dqKObuiabV0aylwUqepcbb70vn0uX9E3TcEMQ+DYF6ze+PutQXGxq6Jv7N3r01ddk9NcPft39vxTmWffSX8tmLz0hnZ2V9X7OY3v9YKCswJ5n8iajKrdS8t3iid1N4xPJ4idz+9UI/96/igJdfo8naVs79xN/JUmzMb3OzHJKeOJCRgo+7l5qzADL7Z944Wn8jOzCk5a8ib/acN7HyfVAndYleYria43e0h7nPWdBZzz/5xf5vgYW5tTOXNqLtSbDm+vthE5nr/+w37jjah77A2mjeDefgt5ntrFg2+98v/vxa2rzFNb07ezWqvL8JGOXdtXgr8hitdaB7KdzJBg2ZnfzevexJl109K+TnE1UN6n7S6vyZr5+bY+vHszofsShZre1nOi7v3AT6P64kZuSJ2PknZf60Q3mwwJ8vz1OnSoaCe8M7uuJcrk7Ei9Kjdz27tsxffzQXKyL9VxPQNRsTIlZ3NUxj9lL4/yd2Ja6Pfoe7+M0hmds3OAhSQOt0XS1ndki2TViko7hxXzNoqgf/cyda+i5gXPoGWoF09gWZ0Zzf/7zM9zHdpHJr7LbaBOfvATZ5Xy3u10q8Z9iypb/Z370bvFliBuTp8j+pz38BT29v6DvWyX2sIwbFKvkPt4ELUHk6+QNGB5N+au1AJ9UDuUuDcV/5rn4ZTR9qfRiP2ADqKymkZXUxj6Ta4iXQv3U+T6AlaT1PpebhDqRruMNoMdzi9Sm/TD2g7fUFz6WuupHk8lA+iX/DH/DWtRvuT9NsXlAePHminhPaBk/+q7oO4fnAdaSCw2Uux6QUcxiF9ArDoh9YPRtphaG0QHQE3hGYA10o6Bm4EnQA3kk6k+Sh5OtxoWgw3hs6GG0vnwh2AXl1IB9ItcAehZ6tQ651wk4Hnapqi/ZxKa2gtHYLePknT0eP1dCT6uhktvQQ3m7bCHUU19Bs6mr6mf9KPQPBSOpnLuIyWcGfuRmdyPx5A5/IQHkpVPJrH0zI+CBT5CU/myXQJz+Cj6FKew3PoCq7iKrqSL+AL6Cq+kq+kq/lWvp+u4TW8hm7mR3kL3cLv8Xu0jv/IDfQ4aPoxPcOf8+e0gb/ir2gjpzhFz4LC+2G8O4LKneE6Une4vUDdfagMdB2JuLGgUQUdR8fTNJpHJ4GKJ9NC0PBUqkLfltHViL2ebqJzlCIX0F1wF4Ii99FF9BA9Am54jNbRZaBLNf07vQha3ABKvEK/BC220c0Y/yQ9QF+BFms4Cips5h58DP2a5/LxzDwPLsQn8dUc5mv5Oh7B18Ptzyv45zyKb4Ebw3fw/TyWH+KHeQo/AncwP84beRq/wwmeyX+HW8RJuMXo+Q4+Q3rOZwq/gwKT0K+J4JYpdArG7gzw2WHg0iWeZMyz4Uk2LMFdf+XJeeCaGfCnglP6gIuGQjZi4Cz5ysDeymXuVQrOIXDjJBqPVifQmXBngY8m6t0k5fTJwEC4NAwqR4DJoeDVw70a5gIr+bXkbHDQ0fq+52z6kcbPBf6HgXMXAv/TaBGNAueORj+WoLy5DoQbh5alpYO1hOumoz7jsu0c77lF1gWv7vBH2vvxqNFcs5F/IPiGwCs/BE1Mq9Lf2aDRDFoAnhE/D5iZ91CSj2iAyp7kHq9+vFLEUGWS4isuF4firklK5bOVVmcrPgdqG+M8N0lbNG6CdeRhkB0lcZSD2SQds8k6bq6bat3BPjfNukN0TMW51J8N+rtuljcSszG+xs217/XMZcZ7lh0l12VHa6G6Uzx3mnWLfG6xdWcof4g7DPx0WICf3Dv/daR3dwB06f7Qsd2gF/ajwdAXvTX+RDhSb66xmBtG0jDqSp2gZfeFft5np8bQvUQrz4cOmm/dcZgBjpMvfAAeB/wqoK1mqjuW5qiX6wfWn6qS1N5KvKPzwSGQ1ApIWxlSest7QGA7DJIjPCnXAvCzXMNBH7k7HLwzAJLfC34QKDQLvf0h6H4SOPko8PVMYNTDYjsZ494JYRf7bMLOXm+6WD8E2mYoaNoVXBHW97kRlS+5ugGrfuhhP/SxHe57Q8cIlFm3H3xP+ApQdjB4cy/wXpndQZLyneC62JbkKzZZ189zva3r64sbbF3wmgapKIEcyMVen6QuuY6AdttbNaJptQvie8F1AG+IH0Bl5ptYmk90hqFqJy3hv7qoG6ouaA8VezkK+yhteis+4tjW3MXD3d9el5w6OlPjK7f8EM+FrBvqc12tC+uYistH/56++3bW/cAblex4u6M0XUfeP1rmucJz+1o32Of2sq5M+UNceUFn+H8sZrMRGG/3Og4aaT7kaQak7FjVSidAA+yPeWYUUkf5qDQMJUV+xsKP2Knx818j4XpiJu1pXdS6SYqrwW8/de2hW9p7+qW7+kFqQ48j+502HlfySuA7bQd432kT+W8H2B1UrkC/xkGWDod0z4F8n+p9v+1w/X7bWfr9tl/o99vW6/fb3rF1yPfb5GtIfdDyKPD4VIzlbGiJ+dDAkiOm33XrpJy5L3hkNGTqYIzpUaDvSdBtJo988a0zZEckUah5EGTvCKX5ApkDNY9YwF1Agf7gvuGg9URosiOhg47H7LrY5pGvxHWFtA0A741AXydBv8zA6J2AGeIMm0e+H9cNVBsICo6EbpsMPTcTGm0e5o4ldOai05Ze4CxVuEzhxQovV3iNwhUKVy46bfkZziqFqxU+oPBRhU8q3KCwenHVeT92tip8TeHvFL6tsE7hhwo/XrLstEXOZwJDUYVDFc5UeI7CFQrXLj37zNNCv1X4lsJ3Ff5RYYPCTxVuX1p14Y9DOxR+KzBMCksUdlTYRWGvpectWhruq3CQwqEKRykchyzLwpMVHqrwSIVHKzxO4XyFp54ntS1RuFThMoUXK7xc4TUKV5y3bHFVeKXCVQpXK3xA4aMKn1S4YTloHq5WuFXhawp/p/BthXUKP1x+dtWS8McKP1P4hcKUwv8WGHEUli5fPmp0pExhN4V7K+yvcLDC4QrHAI6JTFA4VeHhCmcqPFbh8QoXLL/w/OWR0xWepbBK4QUKL1F4hcLrVEbLi4a9ioadi4RlkDv5kk+JfI8NUhrVeWsv1Zid/gXSGVqpWNi9aEg+aL7r6PhiuhQNexQNexYNexcNuxYN9y4a7lM07FYU7A/tPRPz0Xy6kVbRPbSGnsS6/SV6jWrpXfqQPsEKOkX/5BIuw8q5Lw/mkTyOZ/NxvAAr3qV8gRkfnmjDCTY8WmfRnshZxSt5Pb/F252OTqVzuLPQudhZ6TzqvOS843zq/HeoLNQ/NCo0LXR0aIGW4dBkG0634RwbnmrD8214lQ1X2fBxG9bY8EMbfmvCcFjnZQ53MjiGL7PhJTa82IYX2dD2LfyADd+y4btaX2lkaGRqZE5kSeSSyI2RByLPmtTIBhtutuFrptXIB+a55HQbopWSlfQNOYAdeRH9mfvQ56B0OWg8gQ/jU0Ddy5HnRv1aa4eSVXncSq1DrpDWWYV8F+RxVZojWjKnkZuNUrN9NQxHrjGN3HBNb1/SJcd1RImO2dKR7dQ+siPHbdfU0si7AQdawmdLbkCO6oDboGntIvf43B3IeYev1OVIv8bnLteUksg5nluMXIt9JWYjdY7nZmt8JDLRugOQ4wBf7r5IG2RdX40NR8rUlSK1NJszvIPC4W/V7dC4UPgTuI+Q8pEvVy3i34Gr1Rgn/EoYXAKfzfE4Yp8GlGcOr1afTb1RvUm7Dv4KE5Ls8Zocyzy+dmNORY3nh88JxM0GHgvgjg/EjkMfpqubFogHFcIjrasMpISpJNzNc2X+tNCn1C70jc99GUj9LZWGPgi4dwLpa6l9qDrHbQjkuJ6ioTsauZu9PByC9IQWk5k3+mnMHHhQJ3S6jeNQFea2TrBzp8MKPguSGsXMsbfuNoVoEz/PN2r4gs6Dm7iab9DwJZQP8002zXx/OGZz/NKXY6Xv/mb33vkt3wJYwz8DfInvBnyeHyHH+QK2RU/aTLfR/VgtDNLV1hBY2fvBph8Ge7wXvYqeyS7581gHmF1xx6mhXvwwr+F1fCs/wA/xHXwn/4rv4tV8N9/D9/J9fD8/xo/yWl7Ft/MjfBs/iFKvoK8P0IPqHWcrDePPeTt/wl9xiv/OO/gLTvLX/CV/zH/hv3GC4/wp/5U/azF+oLqzFnQuNaE7Es4q+B2YJ36TjWOMrwOJca7yxd2H59+Z0Iv7BTy0La/21fcknqGn+XpfPsg+xo+dh3z5IC0Mfe+stHGOI99qepfr+H3+gP+A5zuohD/i9/iPXM8f8p8Qs4o6IOZdxNlcXho7t5JZnfUHHUZh3TVNV7LzsBZaTOfQ+XQRXUZX8VVo+0Say1dreBJfo+ECvlbDk/k6DRfyTzU8hf8sIdq7Arw1l68EPBE9c1DyZ4ALeAXgyfxzwIWghoMSDeTwe8BV7LNN/Dg/wU/yf2DefYqf5md4A29E+jdUxv/AGD/Lz4GzXwCvbuLN/CK/xFtkfNXWWkCyX/8zupvepL/Qt5jzB/J4PphP4rP438QO4wo+W3MO1C+LyxsaY6VtkmfZGyfSL41vgiO0tUVTxdJhUNrg9RQ/o7lv5XVaupdyTQ9wVy/ddWjMW7InMFo573lw4KsiUVh7HooV7umoSXYAxebqD+cAs4Gg31i4MNa38g28KoxEKWybR5DzMVoL6X6ansGoyfuCzopFV5SLUTFvqeQ9VIm+h+qv7yGG6Buo/fTd0wiqobex4pW3TlP0rdM0fesklkQI+aVV6fcdfL/2ey9rSUNy+Aa+kX8JW+kmvhly/KBiMJ9n8izYUEfhqQQYnMIn8Dw+kefjuQOe5yPG5vClMHjI0REarmMR1h0jQx0zYo5i0pHvRfIA8Psb1IcfAobvUTteQ3vzidAp71N7vodKuY4O4bdpPG+mctgjvZ2nqYtzI/yLWDN9S/P4OdoPcjMVtiPBfjkSFuAwfpm68dPU13mD9nFkt+wJxPVSH+FDqbeUoW00mjuBmnU0lYeBc3vSUO5PPfgx2gc49OQPIHUPA4+/oty1GKXtNAj3g+kjOgR+NK+iIfwqjUU4hk+jHqFONIIzKPMZ8P2A9nUGInwZ/jE6xOmO59Eo9z7BrqLuvAPxDyD+YfiR8BNoX34X4USEcaQJDV6kMmcQcPuaOjh/oE6Qnm5KE6FDP+SpBE790JcraCB3z/wTI3oC34m0t2kcNEwfhP3AV324A8X0y2lv0jh6mcbT1swboLvcT3CeQ17EQwv20XJvoUx/hEL/gSj3azoAnNDBqaK90M+9dDwwBjwFfPMV2puC++Opr/ZbPPotfXb7pPgLTvm84Lgu6IET8Mr8Ff5L+P8Cbvsrbrle8PL79cDhh1TJP0Hai8DhQYzjGIzPDjzXoz/nUVfnHhroDAUd76dHMSYLnYZMBjiWQtf15GWg5d1U7twHvH4DHjwdvCW8tDjzMHRdBXTcIOc8w2NuX6V/oQj6+DHaOBO8Ngc4TAH9pkBfPAPN/hQNV94UmmHc6I1MFX+CFc92SIDw0rM0ACucqcg7DGWGhnuAllK38Iwbvmx4ht7P/A/CEHyp0tX1oK/raStk5H46Dn4W/LnwC+HXwF8Kvxb+JPiL4OfyXXQ7/Ab4u/H8Ovw58Cfb8An4E62XOh6xdcyyz6+At0t0jLsDN+En4bVaw18eL7h0Ap2VL3K8Jx85XmXF7w+imCs7oEMaYTl8R9x/prL0tZEl9Q9auRqZ4/00M75U5cz1Im+5HvLn935aBzxk0+9VRl05dX0O7yufCf/neldmG/towL8JPlR5zqRA807whPu4J9+rjXyrfxX6NkUTVM4/yjRo+KZP7l3/uobtVP6Nr1Q9kOtFL/h9rixaLzrD71UOXP1hvfN7lP8HxUJb4TdjnJ8wPvRzhE/Cv4X0V4x37rXhg9D9j8C/A93+Bmj5J9rbWYe6YBPy7dDdm6AnbwJ9H8EYfEoDuJy6wmrpybch/AB13IrwNejMq6Ej3kb4a5Stpd7hdsDpXdBJ/Id0MP8e4X/CPwVaPEVTIK89eB54722McT/QoTfaOAj3vRE/Efc90cZEvf/u8h0BK+UZzENvwYr5C2QRod9LnOe/BY/8DXK8DTSJg67Qj7BY91G91x9j+BMa4iyGdeHqqxVIX4GwAjp5IHiwP+SmP2g2DbhMQ1hJXXgIePEbmu88jnlyJtH/ArfDUkIKZW5kc3RyZWFtCmVuZG9iagoyMSAwIG9iago1MDkwMQplbmRvYmoKMjAgMCBvYmoKMTI2NDY0CmVuZG9iagoxNiAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDkxID4+CnN0cmVhbQp4nGNgoBAIA7EomCUBJqUZZOByikCszKDCoAqk1YFYE4i1sZpihsS2IdEFIQyhQDKcIYIhkiEKyIphiAWS8QwJDIkMSQzJQHYqQxpDOkMGQyZDFpCXAwDAkAl0CmVuZHN0cmVhbQplbmRvYmoKMTkgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAzNzQgPj4Kc3RyZWFtCnicXZNPb4MwDMXvfIoct0NFoTRZJYQ0dRcO+6OxnaodKDEV0ggo0EO//ZI8Q6Uhtdb7+bl2UhMfy5fSdLOIP+zQVDSLtjPa0jRcbUPiTJfOREkqdNfMrMJ309djFLvi6jbN1JemHaI8F/GnS06zvYmHZz2c6TESQsTvVpPtzEU8fB8roOo6jr/Uk5nFNioKoal1P/daj291TyIOxZtSu3w33zau7O74uo0k0qATjNQMmqaxbsjW5kJRvnVPIfLWPUVERv/LJwpl53b1p96PcEL8CTgFThmnjPfAe8Z7xk/AB8arRJYgibMEvEPnHbdYZLJQmDLIjE0ZYwksGUvGaOzDCTHgbBewDydE4AMwT53xuBITSJ7rLpHFQFJxliXuRcolCS8ayJq9q0S2CVLxzbOUGgF3JtvFE0oU/hXFzReJ4ymMojCKkosVleit+KzKn9Xtx7IIflX8Xq972FytdSsYlj/snt+6ztD6fozD6Kv85w+zktm4CmVuZHN0cmVhbQplbmRvYmoKMTQgMCBvYmoKPDwgL0Jhc2VGb250IC9HdWFyZGlhblNhbnNDb25kLVJlZ3VsYXIKL0NJRFN5c3RlbUluZm8gPDwgL09yZGVyaW5nIChJZGVudGl0eSkgL1JlZ2lzdHJ5IChBZG9iZSkgL1N1cHBsZW1lbnQgMCA+PgovQ0lEVG9HSURNYXAgMTYgMCBSIC9Gb250RGVzY3JpcHRvciAxMyAwIFIgL1N1YnR5cGUgL0NJREZvbnRUeXBlMgovVHlwZSAvRm9udCAvVyAxOCAwIFIgPj4KZW5kb2JqCjE1IDAgb2JqCjw8IC9CYXNlRm9udCAvR3VhcmRpYW5TYW5zQ29uZC1SZWd1bGFyIC9EZXNjZW5kYW50Rm9udHMgWyAxNCAwIFIgXQovRW5jb2RpbmcgL0lkZW50aXR5LUggL1N1YnR5cGUgL1R5cGUwIC9Ub1VuaWNvZGUgMTkgMCBSIC9UeXBlIC9Gb250ID4+CmVuZG9iagoxMyAwIG9iago8PCAvQXNjZW50IDgwOSAvQ2FwSGVpZ2h0IDAgL0Rlc2NlbnQgLTE5MSAvRmxhZ3MgMzIKL0ZvbnRCQm94IFsgLTE0NiAtMjE2IDEwMTggMTA0NCBdIC9Gb250RmlsZTIgMTcgMCBSCi9Gb250TmFtZSAvR3VhcmRpYW5TYW5zQ29uZC1SZWd1bGFyIC9JdGFsaWNBbmdsZSAwIC9NYXhXaWR0aCA3MTUgL1N0ZW1WIDAKL1R5cGUgL0ZvbnREZXNjcmlwdG9yIC9YSGVpZ2h0IDAgPj4KZW5kb2JqCjE4IDAgb2JqClsgMzIgWyAxNTggXSAzNCBbIDMxMiBdIDM3IFsgNzE1IF0gNDAgWyAyODAgMjgwIF0gNDYgWyAyMTYgXSA0OApbIDUzNSAyODggNDI1IF0gNTIgWyA0NzQgXSA1NCBbIDQ4OSBdIDU2IFsgNDk0IF0gNjcgWyA0NzQgXSA3MyBbIDIxNyBdIDk3ClsgNDA5IDQ2NiBdIDEwMCBbIDQ2NiA0MjcgMjY3IDQyNyBdIDEwNSBbIDE5MyAxOTIgXSAxMDgKWyAxOTMgNjk1IDQ2MSA0NTYgNDY2IF0gMTE0IFsgMjg4IDM0NSAyOTAgNDU1IDQxOSA2MjIgXSAxMjEgWyA0MTQgXSBdCmVuZG9iagozIDAgb2JqCjw8IC9GMSAxNSAwIFIgPj4KZW5kb2JqCjQgMCBvYmoKPDwgL0ExIDw8IC9DQSAwIC9UeXBlIC9FeHRHU3RhdGUgL2NhIDEgPj4KL0EyIDw8IC9DQSAxIC9UeXBlIC9FeHRHU3RhdGUgL2NhIDEgPj4gPj4KZW5kb2JqCjUgMCBvYmoKPDwgPj4KZW5kb2JqCjYgMCBvYmoKPDwgPj4KZW5kb2JqCjcgMCBvYmoKPDwgPj4KZW5kb2JqCjIgMCBvYmoKPDwgL0NvdW50IDEgL0tpZHMgWyAxMSAwIFIgXSAvVHlwZSAvUGFnZXMgPj4KZW5kb2JqCjIyIDAgb2JqCjw8IC9DcmVhdGlvbkRhdGUgKEQ6MjAyMTEwMjMxNzEwMDUrMDInMDAnKQovQ3JlYXRvciAoTWF0cGxvdGxpYiB2My40LjMsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcpCi9Qcm9kdWNlciAoTWF0cGxvdGxpYiBwZGYgYmFja2VuZCB2My40LjMpID4+CmVuZG9iagp4cmVmCjAgMjMKMDAwMDAwMDAwMCA2NTUzNSBmIAowMDAwMDAwMDE2IDAwMDAwIG4gCjAwMDAwNTQxNjkgMDAwMDAgbiAKMDAwMDA1Mzk3NSAwMDAwMCBuIAowMDAwMDU0MDA3IDAwMDAwIG4gCjAwMDAwNTQxMDYgMDAwMDAgbiAKMDAwMDA1NDEyNyAwMDAwMCBuIAowMDAwMDU0MTQ4IDAwMDAwIG4gCjAwMDAwMDAwNjUgMDAwMDAgbiAKMDAwMDAwMDQwMSAwMDAwMCBuIAowMDAwMDAxNDIzIDAwMDAwIG4gCjAwMDAwMDAyMDggMDAwMDAgbiAKMDAwMDAwMTQwMyAwMDAwMCBuIAowMDAwMDUzNDY1IDAwMDAwIG4gCjAwMDAwNTMwOTEgMDAwMDAgbiAKMDAwMDA1MzMxMSAwMDAwMCBuIAowMDAwMDUyNDgxIDAwMDAwIG4gCjAwMDAwMDE0NDMgMDAwMDAgbiAKMDAwMDA1MzY5NSAwMDAwMCBuIAowMDAwMDUyNjQ0IDAwMDAwIG4gCjAwMDAwNTI0NTggMDAwMDAgbiAKMDAwMDA1MjQzNiAwMDAwMCBuIAowMDAwMDU0MjI5IDAwMDAwIG4gCnRyYWlsZXIKPDwgL0luZm8gMjIgMCBSIC9Sb290IDEgMCBSIC9TaXplIDIzID4+CnN0YXJ0eHJlZgo1NDM4NgolJUVPRgo=\n", + "image/svg+xml": "\n\n\n \n \n \n \n 2021-10-23T17:10:05.056089\n image/svg+xml\n \n \n Matplotlib v3.4.3, https://matplotlib.org/\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "preds_df = pd.DataFrame(preds[0])\n", + "plt.bar(labels, 100 * preds_df[\"score\"], color='C0')\n", + "plt.title(f'\"{custom_tweet}\"')\n", + "plt.ylabel(\"Class probability (%)\")\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Conclusion" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Congratulations, you now know how to train a transformer model to classify the emotions in tweets! We have seen two complementary approaches based on features and fine-tuning, and investigated their strengths and weaknesses. \n", + "\n", + "However, this is just the first step in building a real-world application with transformer models, and we have a lot more ground to cover. Here's a list of challenges you're likely to experience in your NLP journey:\n", + "\n", + "My boss wants my model in production yesterday!::\n", + "In most applications, your model doesn't just sit somewhere gathering dust - you want to make sure it's serving predictions! When a model is pushed to the Hub, an inference endpoint is automatically created that can be called with HTTP requests. We recommend checking out the [documentation](https://api-inference.huggingface.co/docs/python/html/index.html) of the Inference API if you want to learn more. \n", + "\n", + "My users want faster predictions!::\n", + "We've already seen one approach to this problem: using DistilBERT. In <> we'll dive into knowledge distillation (the process by which DistilBERT was created), along with other tricks to speed up your transformer models.\n", + "\n", + "\n", + "Can your model also do X?::\n", + "As we've alluded to in this chapter, transformers are extremely versatile. In the rest of the book we will be exploring a range of tasks, like question answering and named entity recognition, all using the same basic architecture.\n", + "\n", + "None of my texts are in English!::\n", + "It turns out that transformers also come in a multilingual variety, and we'll use them in <> to tackle several languages at once.\n", + "\n", + "I don't have any labels!::\n", + "If there is very little labeled data available, fine-tuning may not be an option. In <>, we'll explore some techniques to deal with this situation.\n", + "\n", + "Now that we've seen what's involved in training and sharing a transformer, in the next chapter we'll explore implementing our very own transformer model from scratch." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "book", + "language": "python", + "name": "python3" + } + }, + "nbformat": 4, + "nbformat_minor": 4 +}