Skip to content

Commit

Permalink
Make 0.2.1 Release + Update docs (#116)
Browse files Browse the repository at this point in the history 
* Update version from 0.2.0 (yanked) to 0.2.1

* Update text embedding to include prefix for passages and queries

* Update supported models to use the latest API

* * fix(text_embedding_base.py): remove unnecessary prefix from texts in embed method
* feat(text_embedding_base.py): update query_embed method to updated instruction for the v1.5 model

* Remove comparison, since the ranking is identical even with varying embedding

* Refactor text embedding query handling
  • Loading branch information
@NirantK
NirantK authored Feb 8, 2024
1 parent 973da35 commit 81bab0c
Show file tree
Hide file tree
Showing 3 changed files with 119 additions and 122 deletions.
76 changes: 23 additions & 53 deletions docs/examples/Retrieval_with_FastEmbed.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -39,11 +39,19 @@
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[32m2024-02-07 22:20:57.013\u001b[0m | \u001b[33m\u001b[1mWARNING \u001b[0m | \u001b[36mfastembed.embedding\u001b[0m:\u001b[36m<module>\u001b[0m:\u001b[36m7\u001b[0m - \u001b[33m\u001b[1mDefaultEmbedding, FlagEmbedding, JinaEmbedding are deprecated. Use TextEmbedding instead.\u001b[0m\n"
]
}
],
"source": [
"from typing import List\n",
"import numpy as np\n",
"from fastembed.embedding import FlagEmbedding as Embedding"
"from fastembed import TextEmbedding"
]
},
{
Expand Down Expand Up @@ -84,7 +92,7 @@
" \"His life has been depicted in various films, TV shows, and books\",\n",
"]\n",
"# Initialize the DefaultEmbedding class with the desired parameters\n",
"embedding_model = Embedding(model_name=\"BAAI/bge-small-en\", max_length=512)\n",
"embedding_model = TextEmbedding(model_name=\"BAAI/bge-small-en\", max_length=512)\n",
"\n",
"# We'll use the passage_embed method to get the embeddings for the documents\n",
"embeddings: List[np.ndarray] = list(\n",
Expand Down Expand Up @@ -124,65 +132,27 @@
" print(f\"Rank {i+1}: {documents[sorted_scores[i]]}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Running and Comparing Queries\n",
"Finally, we run our sample query using the `print_top_k` function.\n",
"\n",
"The differences between using query embeddings and plain embeddings can be observed in the retrieved ranks:\n",
"\n",
"Using query embeddings (from `query_embed` method):"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Rank 1: Maharana Pratap was a Rajput warrior king from Mewar\n",
"Rank 2: Maharana Pratap is considered a symbol of Rajput resistance against foreign rule\n",
"Rank 3: His legacy is celebrated in Rajasthan through festivals and monuments\n",
"Rank 4: He had 11 wives and 17 sons, including Amar Singh I who succeeded him as ruler of Mewar\n",
"Rank 5: He fought against the Mughal Empire led by Akbar\n"
]
}
],
"source": [
"print_top_k(query_embedding, embeddings, documents)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Using plain embeddings (from `embed` method):"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Rank 1: Maharana Pratap was a Rajput warrior king from Mewar\n",
"Rank 2: Maharana Pratap is considered a symbol of Rajput resistance against foreign rule\n",
"Rank 3: His legacy is celebrated in Rajasthan through festivals and monuments\n",
"Rank 4: He had 11 wives and 17 sons, including Amar Singh I who succeeded him as ruler of Mewar\n",
"Rank 5: He fought against the Mughal Empire led by Akbar\n"
]
"data": {
"text/plain": [
"(array([-0.04393955, 0.04452892, -0.00760788, -0.03399807, 0.01951348],\n",
" dtype=float32),\n",
" array([-0.06002192, 0.04322132, -0.00545516, -0.04419701, -0.00542277],\n",
" dtype=float32))"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"print_top_k(plain_query_embedding, embeddings, documents)"
"query_embedding[:5], plain_query_embedding[:5]"
]
},
{
Expand Down
163 changes: 95 additions & 68 deletions docs/examples/Supported_Models.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
"cells": [
{
"cell_type": "code",
"execution_count": null,
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
Expand All @@ -12,7 +12,7 @@
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 4,
"metadata": {},
"outputs": [
{
Expand Down Expand Up @@ -40,6 +40,7 @@
" <th>dim</th>\n",
" <th>description</th>\n",
" <th>size_in_GB</th>\n",
" <th>sources</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
Expand All @@ -49,143 +50,169 @@
" <td>768</td>\n",
" <td>Base English model</td>\n",
" <td>0.50</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en.tar.gz'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>BAAI/bge-base-en-v1.5</td>\n",
" <td>768</td>\n",
" <td>Base English model, v1.5</td>\n",
" <td>0.44</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en-v1.5.tar.gz', 'hf': 'qdrant/bge-base-en-v1.5-onnx-q'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>BAAI/bge-large-en-v1.5</td>\n",
" <td>BAAI/bge-large-en-v1.5-quantized</td>\n",
" <td>1024</td>\n",
" <td>Large English model, v1.5</td>\n",
" <td>1.34</td>\n",
" <td>{'hf': 'qdrant/bge-large-en-v1.5-onnx-q'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>BAAI/bge-large-en-v1.5</td>\n",
" <td>1024</td>\n",
" <td>Large English model, v1.5</td>\n",
" <td>1.34</td>\n",
" <td>{'hf': 'qdrant/bge-large-en-v1.5-onnx'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>BAAI/bge-small-en</td>\n",
" <td>384</td>\n",
" <td>Fast English model</td>\n",
" <td>0.20</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/BAAI-bge-small-en.tar.gz'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <th>5</th>\n",
" <td>BAAI/bge-small-en-v1.5</td>\n",
" <td>384</td>\n",
" <td>Fast and Default English model</td>\n",
" <td>0.13</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-small-en-v1.5.tar.gz', 'hf': 'qdrant/bge-small-en-v1.5-onnx-q'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>5</th>\n",
" <th>6</th>\n",
" <td>BAAI/bge-small-zh-v1.5</td>\n",
" <td>512</td>\n",
" <td>Fast and recommended Chinese model</td>\n",
" <td>0.10</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-small-zh-v1.5.tar.gz'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>6</th>\n",
" <th>7</th>\n",
" <td>sentence-transformers/all-MiniLM-L6-v2</td>\n",
" <td>384</td>\n",
" <td>Sentence Transformer model, MiniLM-L6-v2</td>\n",
" <td>0.09</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/sentence-transformers-all-MiniLM-L6-v2.tar.gz', 'hf': 'qdrant/all-MiniLM-L6-v2-onnx'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>8</th>\n",
" <td>intfloat/multilingual-e5-large</td>\n",
" <td>1024</td>\n",
" <td>Multilingual model, e5-large. Recommend using this model for non-English languages</td>\n",
" <td>2.24</td>\n",
" <td>{'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-multilingual-e5-large.tar.gz', 'hf': 'qdrant/multilingual-e5-large-onnx'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>7</th>\n",
" <th>9</th>\n",
" <td>sentence-transformers/paraphrase-multilingual-mpnet-base-v2</td>\n",
" <td>768</td>\n",
" <td>Sentence-transformers model for tasks like clustering or semantic search</td>\n",
" <td>1.11</td>\n",
" <td>{'hf': 'xenova/paraphrase-multilingual-mpnet-base-v2'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10</th>\n",
" <td>jinaai/jina-embeddings-v2-base-en</td>\n",
" <td>768</td>\n",
" <td>English embedding model supporting 8192 sequence length</td>\n",
" <td>0.55</td>\n",
" <td>{'hf': 'xenova/jina-embeddings-v2-base-en'}</td>\n",
" </tr>\n",
" <tr>\n",
" <th>8</th>\n",
" <th>11</th>\n",
" <td>jinaai/jina-embeddings-v2-small-en</td>\n",
" <td>512</td>\n",
" <td>English embedding model supporting 8192 sequence length</td>\n",
" <td>0.13</td>\n",
" </tr>\n",
" <tr>\n",
" <th>9</th>\n",
" <td>sentence-transformers/all-MiniLM-L6-v2</td>\n",
" <td>384</td>\n",
" <td>Sentence Transformer model, MiniLM-L6-v2</td>\n",
" <td>0.09</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10</th>\n",
" <td>xenova/multilingual-e5-large</td>\n",
" <td>1024</td>\n",
" <td>Multilingual model. Recommended for non-English languages</td>\n",
" <td>2.24</td>\n",
" </tr>\n",
" <tr>\n",
" <th>11</th>\n",
" <td>xenova/paraphrase-multilingual-mpnet-base-v2</td>\n",
" <td>768</td>\n",
" <td>Sentence-transformers model for tasks like clustering or semantic search</td>\n",
" <td>1.11</td>\n",
" <td>{'hf': 'xenova/jina-embeddings-v2-small-en'}</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" model dim \\\n",
"0 BAAI/bge-base-en 768 \n",
"1 BAAI/bge-base-en-v1.5 768 \n",
"2 BAAI/bge-large-en-v1.5 1024 \n",
"3 BAAI/bge-small-en 384 \n",
"4 BAAI/bge-small-en-v1.5 384 \n",
"5 BAAI/bge-small-zh-v1.5 512 \n",
"6 intfloat/multilingual-e5-large 1024 \n",
"7 jinaai/jina-embeddings-v2-base-en 768 \n",
"8 jinaai/jina-embeddings-v2-small-en 512 \n",
"9 sentence-transformers/all-MiniLM-L6-v2 384 \n",
"10 xenova/multilingual-e5-large 1024 \n",
"11 xenova/paraphrase-multilingual-mpnet-base-v2 768 \n",
" model dim \\\n",
"0 BAAI/bge-base-en 768 \n",
"1 BAAI/bge-base-en-v1.5 768 \n",
"2 BAAI/bge-large-en-v1.5-quantized 1024 \n",
"3 BAAI/bge-large-en-v1.5 1024 \n",
"4 BAAI/bge-small-en 384 \n",
"5 BAAI/bge-small-en-v1.5 384 \n",
"6 BAAI/bge-small-zh-v1.5 512 \n",
"7 sentence-transformers/all-MiniLM-L6-v2 384 \n",
"8 intfloat/multilingual-e5-large 1024 \n",
"9 sentence-transformers/paraphrase-multilingual-mpnet-base-v2 768 \n",
"10 jinaai/jina-embeddings-v2-base-en 768 \n",
"11 jinaai/jina-embeddings-v2-small-en 512 \n",
"\n",
" description \\\n",
"0 Base English model \n",
"1 Base English model, v1.5 \n",
"2 Large English model, v1.5 \n",
"3 Fast English model \n",
"4 Fast and Default English model \n",
"5 Fast and recommended Chinese model \n",
"6 Multilingual model, e5-large. Recommend using this model for non-English languages \n",
"7 English embedding model supporting 8192 sequence length \n",
"8 English embedding model supporting 8192 sequence length \n",
"9 Sentence Transformer model, MiniLM-L6-v2 \n",
"10 Multilingual model. Recommended for non-English languages \n",
"11 Sentence-transformers model for tasks like clustering or semantic search \n",
"3 Large English model, v1.5 \n",
"4 Fast English model \n",
"5 Fast and Default English model \n",
"6 Fast and recommended Chinese model \n",
"7 Sentence Transformer model, MiniLM-L6-v2 \n",
"8 Multilingual model, e5-large. Recommend using this model for non-English languages \n",
"9 Sentence-transformers model for tasks like clustering or semantic search \n",
"10 English embedding model supporting 8192 sequence length \n",
"11 English embedding model supporting 8192 sequence length \n",
"\n",
" size_in_GB \\\n",
"0 0.50 \n",
"1 0.44 \n",
"2 1.34 \n",
"3 1.34 \n",
"4 0.20 \n",
"5 0.13 \n",
"6 0.10 \n",
"7 0.09 \n",
"8 2.24 \n",
"9 1.11 \n",
"10 0.55 \n",
"11 0.13 \n",
"\n",
" size_in_GB \n",
"0 0.50 \n",
"1 0.44 \n",
"2 1.34 \n",
"3 0.20 \n",
"4 0.13 \n",
"5 0.10 \n",
"6 2.24 \n",
"7 0.55 \n",
"8 0.13 \n",
"9 0.09 \n",
"10 2.24 \n",
"11 1.11 "
" sources \n",
"0 {'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en.tar.gz'} \n",
"1 {'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en-v1.5.tar.gz', 'hf': 'qdrant/bge-base-en-v1.5-onnx-q'} \n",
"2 {'hf': 'qdrant/bge-large-en-v1.5-onnx-q'} \n",
"3 {'hf': 'qdrant/bge-large-en-v1.5-onnx'} \n",
"4 {'url': 'https://storage.googleapis.com/qdrant-fastembed/BAAI-bge-small-en.tar.gz'} \n",
"5 {'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-small-en-v1.5.tar.gz', 'hf': 'qdrant/bge-small-en-v1.5-onnx-q'} \n",
"6 {'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-bge-small-zh-v1.5.tar.gz'} \n",
"7 {'url': 'https://storage.googleapis.com/qdrant-fastembed/sentence-transformers-all-MiniLM-L6-v2.tar.gz', 'hf': 'qdrant/all-MiniLM-L6-v2-onnx'} \n",
"8 {'url': 'https://storage.googleapis.com/qdrant-fastembed/fast-multilingual-e5-large.tar.gz', 'hf': 'qdrant/multilingual-e5-large-onnx'} \n",
"9 {'hf': 'xenova/paraphrase-multilingual-mpnet-base-v2'} \n",
"10 {'hf': 'xenova/jina-embeddings-v2-base-en'} \n",
"11 {'hf': 'xenova/jina-embeddings-v2-small-en'} "
]
},
"execution_count": 6,
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"from fastembed.embedding import Embedding\n",
"from fastembed import TextEmbedding\n",
"import pandas as pd\n",
"\n",
"pd.set_option(\"display.max_colwidth\", None)\n",
"pd.DataFrame(Embedding.list_supported_models())"
"pd.DataFrame(TextEmbedding.list_supported_models())"
]
}
],
Expand All @@ -205,7 +232,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.7"
"version": "3.11.5"
},
"orig_nbformat": 4
},
Expand Down
2 changes: 1 addition & 1 deletion pyproject.toml
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
[tool.poetry]
name = "fastembed"
version = "0.2.0"
version = "0.2.1"
description = "Fast, light, accurate library built for retrieval embedding generation"
authors = ["NirantK <[email protected]>"]
license = "Apache License"
Expand Down

0 comments on commit 81bab0c

Please sign in to comment.