ch03-word2vec.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# cbow模型(Continuous Bag-Of-Words Model)预测\n",
    "\n",
    "https://github.com/oreilly-japan/deep-learning-from-scratch-2/blob/master/ch03/cbow_predict.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.28830492  0.58316349 -0.29270063 -0.05973828  0.6979968  -0.3119117\n",
      "   0.39784004]]\n"
     ]
    }
   ],
   "source": [
    "# coding: utf-8\n",
    "import sys\n",
    "sys.path.append('..')\n",
    "import numpy as np\n",
    "from common.layers import MatMul\n",
    "\n",
    "\n",
    "# 示例上下文数据\n",
    "c0 = np.array([[1, 0, 0, 0, 0, 0, 0]])\n",
    "c1 = np.array([[0, 0, 1, 0, 0, 0, 0]])\n",
    "\n",
    "# 权重初始化\n",
    "W_in = np.random.randn(7, 3)\n",
    "W_out = np.random.randn(3, 7)\n",
    "\n",
    "# 生成层\n",
    "in_layer0 = MatMul(W_in)\n",
    "in_layer1 = MatMul(W_in)\n",
    "out_layer = MatMul(W_out)\n",
    "\n",
    "# 前向传播\n",
    "h0 = in_layer0.forward(c0)\n",
    "h1 = in_layer1.forward(c1)\n",
    "h = 0.5 * (h0 + h1)\n",
    "s = out_layer.forward(h)\n",
    "print(s)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# simple cbow模型实现\n",
    "\n",
    "https://github.com/oreilly-japan/deep-learning-from-scratch-2/blob/master/ch03/simple_cbow.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# coding: utf-8\n",
    "import sys\n",
    "sys.path.append('..')\n",
    "import numpy as np\n",
    "from common.layers import MatMul, SoftmaxWithLoss\n",
    "\n",
    "\n",
    "class SimpleCBOW:\n",
    "    def __init__(self, vocab_size, hidden_size):\n",
    "        V, H = vocab_size, hidden_size\n",
    "\n",
    "        # 权重初始化\n",
    "        W_in = 0.01 * np.random.randn(V, H).astype('f')\n",
    "        W_out = 0.01 * np.random.randn(H, V).astype('f')\n",
    "\n",
    "        # 生成层\n",
    "        self.in_layer0 = MatMul(W_in)\n",
    "        self.in_layer1 = MatMul(W_in)\n",
    "        self.out_layer = MatMul(W_out)\n",
    "        self.loss_layer = SoftmaxWithLoss()\n",
    "\n",
    "        # 将所有的权重和梯度汇总到列表\n",
    "        layers = [self.in_layer0, self.in_layer1, self.out_layer]\n",
    "        self.params, self.grads = [], []\n",
    "        for layer in layers:\n",
    "            self.params += layer.params\n",
    "            self.grads += layer.grads\n",
    "\n",
    "        # 设置单词的离散表示成员变量\n",
    "        self.word_vecs = W_in\n",
    "\n",
    "    def forward(self, contexts, target):\n",
    "        h0 = self.in_layer0.forward(contexts[:, 0])\n",
    "        h1 = self.in_layer1.forward(contexts[:, 1])\n",
    "        h = (h0 + h1) * 0.5\n",
    "        score = self.out_layer.forward(h)\n",
    "        loss = self.loss_layer.forward(score, target)\n",
    "        return loss\n",
    "\n",
    "    def backward(self, dout=1):\n",
    "        ds = self.loss_layer.backward(dout)\n",
    "        da = self.out_layer.backward(ds)\n",
    "        da *= 0.5\n",
    "        self.in_layer1.backward(da)\n",
    "        self.in_layer0.backward(da)\n",
    "        return None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 训练上述模型\n",
    "\n",
    "https://github.com/oreilly-japan/deep-learning-from-scratch-2/blob/master/ch03/train.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 1 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 2 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 3 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 4 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 5 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 6 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 7 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 8 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 9 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 10 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 11 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 12 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 13 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 14 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 15 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 16 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 17 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 18 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 19 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 20 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 21 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 22 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 23 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 24 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 25 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 26 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 27 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 28 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 29 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 30 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 31 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 32 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 33 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 34 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 35 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 36 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 37 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 38 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 39 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 40 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 41 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 42 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 43 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 44 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 45 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 46 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 47 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 48 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 49 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 50 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 51 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 52 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 53 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 54 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 55 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 56 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 57 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 58 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 59 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 60 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 61 |  iter 1 / 2 | time 0[s] | loss 1.86\n",
      "| epoch 62 |  iter 1 / 2 | time 0[s] | loss 1.86\n",
      "| epoch 63 |  iter 1 / 2 | time 0[s] | loss 1.85\n",
      "| epoch 64 |  iter 1 / 2 | time 0[s] | loss 1.85\n",
      "| epoch 65 |  iter 1 / 2 | time 0[s] | loss 1.85\n",
      "| epoch 66 |  iter 1 / 2 | time 0[s] | loss 1.85\n",
      "| epoch 67 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 68 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 69 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 70 |  iter 1 / 2 | time 0[s] | loss 1.83\n",
      "| epoch 71 |  iter 1 / 2 | time 0[s] | loss 1.83\n",
      "| epoch 72 |  iter 1 / 2 | time 0[s] | loss 1.82\n",
      "| epoch 73 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 74 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 75 |  iter 1 / 2 | time 0[s] | loss 1.82\n",
      "| epoch 76 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 77 |  iter 1 / 2 | time 0[s] | loss 1.82\n",
      "| epoch 78 |  iter 1 / 2 | time 0[s] | loss 1.80\n",
      "| epoch 79 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 80 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 81 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 82 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 83 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 84 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 85 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 86 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 87 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 88 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 89 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 90 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 91 |  iter 1 / 2 | time 0[s] | loss 1.74\n",
      "| epoch 92 |  iter 1 / 2 | time 0[s] | loss 1.74\n",
      "| epoch 93 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 94 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 95 |  iter 1 / 2 | time 0[s] | loss 1.74\n",
      "| epoch 96 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 97 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 98 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 99 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 100 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 101 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 102 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 103 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 104 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 105 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 106 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 107 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 108 |  iter 1 / 2 | time 0[s] | loss 1.64\n",
      "| epoch 109 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 110 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 111 |  iter 1 / 2 | time 0[s] | loss 1.64\n",
      "| epoch 112 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 113 |  iter 1 / 2 | time 0[s] | loss 1.63\n",
      "| epoch 114 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 115 |  iter 1 / 2 | time 0[s] | loss 1.62\n",
      "| epoch 116 |  iter 1 / 2 | time 0[s] | loss 1.64\n",
      "| epoch 117 |  iter 1 / 2 | time 0[s] | loss 1.60\n",
      "| epoch 118 |  iter 1 / 2 | time 0[s] | loss 1.64\n",
      "| epoch 119 |  iter 1 / 2 | time 0[s] | loss 1.59\n",
      "| epoch 120 |  iter 1 / 2 | time 0[s] | loss 1.64\n",
      "| epoch 121 |  iter 1 / 2 | time 0[s] | loss 1.59\n",
      "| epoch 122 |  iter 1 / 2 | time 0[s] | loss 1.60\n",
      "| epoch 123 |  iter 1 / 2 | time 0[s] | loss 1.61\n",
      "| epoch 124 |  iter 1 / 2 | time 0[s] | loss 1.59\n",
      "| epoch 125 |  iter 1 / 2 | time 0[s] | loss 1.55\n",
      "| epoch 126 |  iter 1 / 2 | time 0[s] | loss 1.58\n",
      "| epoch 127 |  iter 1 / 2 | time 0[s] | loss 1.59\n",
      "| epoch 128 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 129 |  iter 1 / 2 | time 0[s] | loss 1.61\n",
      "| epoch 130 |  iter 1 / 2 | time 0[s] | loss 1.56\n",
      "| epoch 131 |  iter 1 / 2 | time 0[s] | loss 1.56\n",
      "| epoch 132 |  iter 1 / 2 | time 0[s] | loss 1.53\n",
      "| epoch 133 |  iter 1 / 2 | time 0[s] | loss 1.55\n",
      "| epoch 134 |  iter 1 / 2 | time 0[s] | loss 1.51\n",
      "| epoch 135 |  iter 1 / 2 | time 0[s] | loss 1.55\n",
      "| epoch 136 |  iter 1 / 2 | time 0[s] | loss 1.52\n",
      "| epoch 137 |  iter 1 / 2 | time 0[s] | loss 1.55\n",
      "| epoch 138 |  iter 1 / 2 | time 0[s] | loss 1.51\n",
      "| epoch 139 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 140 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 141 |  iter 1 / 2 | time 0[s] | loss 1.48\n",
      "| epoch 142 |  iter 1 / 2 | time 0[s] | loss 1.53\n",
      "| epoch 143 |  iter 1 / 2 | time 0[s] | loss 1.47\n",
      "| epoch 144 |  iter 1 / 2 | time 0[s] | loss 1.49\n",
      "| epoch 145 |  iter 1 / 2 | time 0[s] | loss 1.48\n",
      "| epoch 146 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 147 |  iter 1 / 2 | time 0[s] | loss 1.46\n",
      "| epoch 148 |  iter 1 / 2 | time 0[s] | loss 1.45\n",
      "| epoch 149 |  iter 1 / 2 | time 0[s] | loss 1.53\n",
      "| epoch 150 |  iter 1 / 2 | time 0[s] | loss 1.39\n",
      "| epoch 151 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 152 |  iter 1 / 2 | time 0[s] | loss 1.46\n",
      "| epoch 153 |  iter 1 / 2 | time 0[s] | loss 1.45\n",
      "| epoch 154 |  iter 1 / 2 | time 0[s] | loss 1.45\n",
      "| epoch 155 |  iter 1 / 2 | time 0[s] | loss 1.43\n",
      "| epoch 156 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 157 |  iter 1 / 2 | time 0[s] | loss 1.43\n",
      "| epoch 158 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 159 |  iter 1 / 2 | time 0[s] | loss 1.42\n",
      "| epoch 160 |  iter 1 / 2 | time 0[s] | loss 1.47\n",
      "| epoch 161 |  iter 1 / 2 | time 0[s] | loss 1.38\n",
      "| epoch 162 |  iter 1 / 2 | time 0[s] | loss 1.43\n",
      "| epoch 163 |  iter 1 / 2 | time 0[s] | loss 1.37\n",
      "| epoch 164 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 165 |  iter 1 / 2 | time 0[s] | loss 1.33\n",
      "| epoch 166 |  iter 1 / 2 | time 0[s] | loss 1.48\n",
      "| epoch 167 |  iter 1 / 2 | time 0[s] | loss 1.30\n",
      "| epoch 168 |  iter 1 / 2 | time 0[s] | loss 1.41\n",
      "| epoch 169 |  iter 1 / 2 | time 0[s] | loss 1.40\n",
      "| epoch 170 |  iter 1 / 2 | time 0[s] | loss 1.39\n",
      "| epoch 171 |  iter 1 / 2 | time 0[s] | loss 1.34\n",
      "| epoch 172 |  iter 1 / 2 | time 0[s] | loss 1.37\n",
      "| epoch 173 |  iter 1 / 2 | time 0[s] | loss 1.36\n",
      "| epoch 174 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 175 |  iter 1 / 2 | time 0[s] | loss 1.28\n",
      "| epoch 176 |  iter 1 / 2 | time 0[s] | loss 1.33\n",
      "| epoch 177 |  iter 1 / 2 | time 0[s] | loss 1.37\n",
      "| epoch 178 |  iter 1 / 2 | time 0[s] | loss 1.38\n",
      "| epoch 179 |  iter 1 / 2 | time 0[s] | loss 1.32\n",
      "| epoch 180 |  iter 1 / 2 | time 0[s] | loss 1.35\n",
      "| epoch 181 |  iter 1 / 2 | time 0[s] | loss 1.36\n",
      "| epoch 182 |  iter 1 / 2 | time 0[s] | loss 1.30\n",
      "| epoch 183 |  iter 1 / 2 | time 0[s] | loss 1.33\n",
      "| epoch 184 |  iter 1 / 2 | time 0[s] | loss 1.33\n",
      "| epoch 185 |  iter 1 / 2 | time 0[s] | loss 1.27\n",
      "| epoch 186 |  iter 1 / 2 | time 0[s] | loss 1.31\n",
      "| epoch 187 |  iter 1 / 2 | time 0[s] | loss 1.34\n",
      "| epoch 188 |  iter 1 / 2 | time 0[s] | loss 1.29\n",
      "| epoch 189 |  iter 1 / 2 | time 0[s] | loss 1.35\n",
      "| epoch 190 |  iter 1 / 2 | time 0[s] | loss 1.24\n",
      "| epoch 191 |  iter 1 / 2 | time 0[s] | loss 1.29\n",
      "| epoch 192 |  iter 1 / 2 | time 0[s] | loss 1.31\n",
      "| epoch 193 |  iter 1 / 2 | time 0[s] | loss 1.24\n",
      "| epoch 194 |  iter 1 / 2 | time 0[s] | loss 1.35\n",
      "| epoch 195 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 196 |  iter 1 / 2 | time 0[s] | loss 1.29\n",
      "| epoch 197 |  iter 1 / 2 | time 0[s] | loss 1.37\n",
      "| epoch 198 |  iter 1 / 2 | time 0[s] | loss 1.21\n",
      "| epoch 199 |  iter 1 / 2 | time 0[s] | loss 1.25\n",
      "| epoch 200 |  iter 1 / 2 | time 0[s] | loss 1.29\n",
      "| epoch 201 |  iter 1 / 2 | time 0[s] | loss 1.22\n",
      "| epoch 202 |  iter 1 / 2 | time 0[s] | loss 1.28\n",
      "| epoch 203 |  iter 1 / 2 | time 0[s] | loss 1.23\n",
      "| epoch 204 |  iter 1 / 2 | time 0[s] | loss 1.30\n",
      "| epoch 205 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 206 |  iter 1 / 2 | time 0[s] | loss 1.26\n",
      "| epoch 207 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 208 |  iter 1 / 2 | time 0[s] | loss 1.25\n",
      "| epoch 209 |  iter 1 / 2 | time 0[s] | loss 1.25\n",
      "| epoch 210 |  iter 1 / 2 | time 0[s] | loss 1.24\n",
      "| epoch 211 |  iter 1 / 2 | time 0[s] | loss 1.22\n",
      "| epoch 212 |  iter 1 / 2 | time 0[s] | loss 1.22\n",
      "| epoch 213 |  iter 1 / 2 | time 0[s] | loss 1.16\n",
      "| epoch 214 |  iter 1 / 2 | time 0[s] | loss 1.26\n",
      "| epoch 215 |  iter 1 / 2 | time 0[s] | loss 1.22\n",
      "| epoch 216 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 217 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 218 |  iter 1 / 2 | time 0[s] | loss 1.21\n",
      "| epoch 219 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 220 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 221 |  iter 1 / 2 | time 0[s] | loss 1.24\n",
      "| epoch 222 |  iter 1 / 2 | time 0[s] | loss 1.17\n",
      "| epoch 223 |  iter 1 / 2 | time 0[s] | loss 1.19\n",
      "| epoch 224 |  iter 1 / 2 | time 0[s] | loss 1.18\n",
      "| epoch 225 |  iter 1 / 2 | time 0[s] | loss 1.18\n",
      "| epoch 226 |  iter 1 / 2 | time 0[s] | loss 1.17\n",
      "| epoch 227 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 228 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 229 |  iter 1 / 2 | time 0[s] | loss 1.29\n",
      "| epoch 230 |  iter 1 / 2 | time 0[s] | loss 1.10\n",
      "| epoch 231 |  iter 1 / 2 | time 0[s] | loss 1.16\n",
      "| epoch 232 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 233 |  iter 1 / 2 | time 0[s] | loss 1.11\n",
      "| epoch 234 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 235 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 236 |  iter 1 / 2 | time 0[s] | loss 1.21\n",
      "| epoch 237 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 238 |  iter 1 / 2 | time 0[s] | loss 1.08\n",
      "| epoch 239 |  iter 1 / 2 | time 0[s] | loss 1.28\n",
      "| epoch 240 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 241 |  iter 1 / 2 | time 0[s] | loss 1.12\n",
      "| epoch 242 |  iter 1 / 2 | time 0[s] | loss 1.02\n",
      "| epoch 243 |  iter 1 / 2 | time 0[s] | loss 1.20\n",
      "| epoch 244 |  iter 1 / 2 | time 0[s] | loss 1.13\n",
      "| epoch 245 |  iter 1 / 2 | time 0[s] | loss 1.18\n",
      "| epoch 246 |  iter 1 / 2 | time 0[s] | loss 1.13\n",
      "| epoch 247 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 248 |  iter 1 / 2 | time 0[s] | loss 1.12\n",
      "| epoch 249 |  iter 1 / 2 | time 0[s] | loss 1.19\n",
      "| epoch 250 |  iter 1 / 2 | time 0[s] | loss 1.04\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 251 |  iter 1 / 2 | time 0[s] | loss 1.11\n",
      "| epoch 252 |  iter 1 / 2 | time 0[s] | loss 1.10\n",
      "| epoch 253 |  iter 1 / 2 | time 0[s] | loss 1.17\n",
      "| epoch 254 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 255 |  iter 1 / 2 | time 0[s] | loss 1.24\n",
      "| epoch 256 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 257 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 258 |  iter 1 / 2 | time 0[s] | loss 1.17\n",
      "| epoch 259 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 260 |  iter 1 / 2 | time 0[s] | loss 1.16\n",
      "| epoch 261 |  iter 1 / 2 | time 0[s] | loss 1.00\n",
      "| epoch 262 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 263 |  iter 1 / 2 | time 0[s] | loss 1.01\n",
      "| epoch 264 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 265 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 266 |  iter 1 / 2 | time 0[s] | loss 1.00\n",
      "| epoch 267 |  iter 1 / 2 | time 0[s] | loss 1.07\n",
      "| epoch 268 |  iter 1 / 2 | time 0[s] | loss 1.07\n",
      "| epoch 269 |  iter 1 / 2 | time 0[s] | loss 1.07\n",
      "| epoch 270 |  iter 1 / 2 | time 0[s] | loss 1.06\n",
      "| epoch 271 |  iter 1 / 2 | time 0[s] | loss 1.06\n",
      "| epoch 272 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 273 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 274 |  iter 1 / 2 | time 0[s] | loss 1.13\n",
      "| epoch 275 |  iter 1 / 2 | time 0[s] | loss 1.13\n",
      "| epoch 276 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 277 |  iter 1 / 2 | time 0[s] | loss 1.12\n",
      "| epoch 278 |  iter 1 / 2 | time 0[s] | loss 1.05\n",
      "| epoch 279 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 280 |  iter 1 / 2 | time 0[s] | loss 0.97\n",
      "| epoch 281 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 282 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 283 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 284 |  iter 1 / 2 | time 0[s] | loss 1.12\n",
      "| epoch 285 |  iter 1 / 2 | time 0[s] | loss 1.03\n",
      "| epoch 286 |  iter 1 / 2 | time 0[s] | loss 1.11\n",
      "| epoch 287 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 288 |  iter 1 / 2 | time 0[s] | loss 1.10\n",
      "| epoch 289 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 290 |  iter 1 / 2 | time 0[s] | loss 1.11\n",
      "| epoch 291 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 292 |  iter 1 / 2 | time 0[s] | loss 1.01\n",
      "| epoch 293 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 294 |  iter 1 / 2 | time 0[s] | loss 1.02\n",
      "| epoch 295 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 296 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 297 |  iter 1 / 2 | time 0[s] | loss 1.08\n",
      "| epoch 298 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 299 |  iter 1 / 2 | time 0[s] | loss 1.00\n",
      "| epoch 300 |  iter 1 / 2 | time 0[s] | loss 1.02\n",
      "| epoch 301 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 302 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 303 |  iter 1 / 2 | time 0[s] | loss 1.08\n",
      "| epoch 304 |  iter 1 / 2 | time 0[s] | loss 1.09\n",
      "| epoch 305 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 306 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 307 |  iter 1 / 2 | time 0[s] | loss 1.06\n",
      "| epoch 308 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 309 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 310 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 311 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 312 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 313 |  iter 1 / 2 | time 0[s] | loss 0.97\n",
      "| epoch 314 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 315 |  iter 1 / 2 | time 0[s] | loss 1.15\n",
      "| epoch 316 |  iter 1 / 2 | time 0[s] | loss 0.96\n",
      "| epoch 317 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 318 |  iter 1 / 2 | time 0[s] | loss 0.97\n",
      "| epoch 319 |  iter 1 / 2 | time 0[s] | loss 1.16\n",
      "| epoch 320 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 321 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 322 |  iter 1 / 2 | time 0[s] | loss 1.03\n",
      "| epoch 323 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 324 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 325 |  iter 1 / 2 | time 0[s] | loss 1.14\n",
      "| epoch 326 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 327 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 328 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 329 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 330 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 331 |  iter 1 / 2 | time 0[s] | loss 1.05\n",
      "| epoch 332 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 333 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 334 |  iter 1 / 2 | time 0[s] | loss 1.01\n",
      "| epoch 335 |  iter 1 / 2 | time 0[s] | loss 1.07\n",
      "| epoch 336 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 337 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 338 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 339 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 340 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 341 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 342 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 343 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 344 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 345 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 346 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 347 |  iter 1 / 2 | time 0[s] | loss 0.93\n",
      "| epoch 348 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 349 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 350 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 351 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 352 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 353 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 354 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 355 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 356 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 357 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 358 |  iter 1 / 2 | time 0[s] | loss 1.05\n",
      "| epoch 359 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 360 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 361 |  iter 1 / 2 | time 0[s] | loss 1.02\n",
      "| epoch 362 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 363 |  iter 1 / 2 | time 0[s] | loss 1.07\n",
      "| epoch 364 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 365 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 366 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 367 |  iter 1 / 2 | time 0[s] | loss 1.01\n",
      "| epoch 368 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 369 |  iter 1 / 2 | time 0[s] | loss 1.04\n",
      "| epoch 370 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 371 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 372 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 373 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 374 |  iter 1 / 2 | time 0[s] | loss 1.01\n",
      "| epoch 375 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 376 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 377 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 378 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 379 |  iter 1 / 2 | time 0[s] | loss 1.08\n",
      "| epoch 380 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 381 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 382 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 383 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 384 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 385 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 386 |  iter 1 / 2 | time 0[s] | loss 0.96\n",
      "| epoch 387 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 388 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 389 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 390 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 391 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 392 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 393 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 394 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 395 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 396 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 397 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 398 |  iter 1 / 2 | time 0[s] | loss 1.03\n",
      "| epoch 399 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 400 |  iter 1 / 2 | time 0[s] | loss 0.93\n",
      "| epoch 401 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 402 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 403 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 404 |  iter 1 / 2 | time 0[s] | loss 0.93\n",
      "| epoch 405 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 406 |  iter 1 / 2 | time 0[s] | loss 0.94\n",
      "| epoch 407 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 408 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 409 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 410 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 411 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 412 |  iter 1 / 2 | time 0[s] | loss 0.97\n",
      "| epoch 413 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 414 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 415 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 416 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 417 |  iter 1 / 2 | time 0[s] | loss 0.96\n",
      "| epoch 418 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 419 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 420 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 421 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 422 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 423 |  iter 1 / 2 | time 0[s] | loss 0.92\n",
      "| epoch 424 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 425 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 426 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 427 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 428 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 429 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 430 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 431 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 432 |  iter 1 / 2 | time 0[s] | loss 0.95\n",
      "| epoch 433 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 434 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 435 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 436 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 437 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 438 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 439 |  iter 1 / 2 | time 0[s] | loss 1.00\n",
      "| epoch 440 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 441 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 442 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 443 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 444 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 445 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 446 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 447 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 448 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 449 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 450 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 451 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 452 |  iter 1 / 2 | time 0[s] | loss 0.93\n",
      "| epoch 453 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 454 |  iter 1 / 2 | time 0[s] | loss 1.02\n",
      "| epoch 455 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 456 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 457 |  iter 1 / 2 | time 0[s] | loss 0.99\n",
      "| epoch 458 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 459 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 460 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 461 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 462 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 463 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 464 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 465 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 466 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 467 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 468 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 469 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 470 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 471 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 472 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 473 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 474 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 475 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 476 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 477 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 478 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 479 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 480 |  iter 1 / 2 | time 0[s] | loss 0.98\n",
      "| epoch 481 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 482 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 483 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 484 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 485 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 486 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 487 |  iter 1 / 2 | time 0[s] | loss 0.89\n",
      "| epoch 488 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 489 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 490 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 491 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 492 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 493 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 494 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 495 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 496 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 497 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 498 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 499 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 500 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 501 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 502 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 503 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 504 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 505 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 506 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 507 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 508 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 509 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 510 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 511 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 512 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 513 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 514 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 515 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 516 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 517 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 518 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 519 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 520 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 521 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 522 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 523 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 524 |  iter 1 / 2 | time 0[s] | loss 0.84\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 525 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 526 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 527 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 528 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 529 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 530 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 531 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 532 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 533 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 534 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 535 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 536 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 537 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 538 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 539 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 540 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 541 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 542 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 543 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 544 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 545 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 546 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 547 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 548 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 549 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 550 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 551 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 552 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 553 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 554 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 555 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 556 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 557 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 558 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 559 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 560 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 561 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 562 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 563 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 564 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 565 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 566 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 567 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 568 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 569 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 570 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 571 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 572 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 573 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 574 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 575 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 576 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 577 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 578 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 579 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 580 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 581 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 582 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 583 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 584 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 585 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 586 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 587 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 588 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 589 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 590 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 591 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 592 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 593 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 594 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 595 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 596 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 597 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 598 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 599 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 600 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 601 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 602 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 603 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 604 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 605 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 606 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 607 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 608 |  iter 1 / 2 | time 0[s] | loss 0.91\n",
      "| epoch 609 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 610 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 611 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 612 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 613 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 614 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 615 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 616 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 617 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 618 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 619 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 620 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 621 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 622 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 623 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 624 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 625 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 626 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 627 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 628 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 629 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 630 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 631 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 632 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 633 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 634 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 635 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 636 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 637 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 638 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 639 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 640 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 641 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 642 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 643 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 644 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 645 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 646 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 647 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 648 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 649 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 650 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 651 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 652 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 653 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 654 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 655 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 656 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 657 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 658 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 659 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 660 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 661 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 662 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 663 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 664 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 665 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 666 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 667 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 668 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 669 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 670 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 671 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 672 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 673 |  iter 1 / 2 | time 0[s] | loss 0.57\n",
      "| epoch 674 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 675 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 676 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 677 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 678 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 679 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 680 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 681 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 682 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 683 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 684 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 685 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 686 |  iter 1 / 2 | time 0[s] | loss 0.90\n",
      "| epoch 687 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 688 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 689 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 690 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 691 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 692 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 693 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 694 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 695 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 696 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 697 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 698 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 699 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 700 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 701 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 702 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 703 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 704 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 705 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 706 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 707 |  iter 1 / 2 | time 0[s] | loss 0.57\n",
      "| epoch 708 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 709 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 710 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 711 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 712 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 713 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 714 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 715 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 716 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 717 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 718 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 719 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 720 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 721 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 722 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 723 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 724 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 725 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 726 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 727 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 728 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 729 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 730 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 731 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 732 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 733 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 734 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 735 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 736 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 737 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 738 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 739 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 740 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 741 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 742 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 743 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 744 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 745 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 746 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 747 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 748 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 749 |  iter 1 / 2 | time 0[s] | loss 0.57\n",
      "| epoch 750 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 751 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 752 |  iter 1 / 2 | time 0[s] | loss 0.57\n",
      "| epoch 753 |  iter 1 / 2 | time 0[s] | loss 0.82\n",
      "| epoch 754 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 755 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 756 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 757 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 758 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 759 |  iter 1 / 2 | time 0[s] | loss 0.88\n",
      "| epoch 760 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 761 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 762 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 763 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 764 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 765 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 766 |  iter 1 / 2 | time 0[s] | loss 0.49\n",
      "| epoch 767 |  iter 1 / 2 | time 0[s] | loss 0.57\n",
      "| epoch 768 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 769 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 770 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 771 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 772 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 773 |  iter 1 / 2 | time 0[s] | loss 0.49\n",
      "| epoch 774 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 775 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 776 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 777 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 778 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 779 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 780 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 781 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 782 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 783 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 784 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 785 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 786 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 787 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 788 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 789 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 790 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 791 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 792 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 793 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 794 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 795 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 796 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 797 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 798 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 799 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 800 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 801 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 802 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 803 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 804 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 805 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 806 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 807 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 808 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 809 |  iter 1 / 2 | time 0[s] | loss 0.80\n",
      "| epoch 810 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 811 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 812 |  iter 1 / 2 | time 0[s] | loss 0.55\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 813 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 814 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 815 |  iter 1 / 2 | time 0[s] | loss 0.45\n",
      "| epoch 816 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 817 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 818 |  iter 1 / 2 | time 0[s] | loss 0.87\n",
      "| epoch 819 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 820 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 821 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 822 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 823 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 824 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 825 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 826 |  iter 1 / 2 | time 0[s] | loss 0.79\n",
      "| epoch 827 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 828 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 829 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 830 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 831 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 832 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 833 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 834 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 835 |  iter 1 / 2 | time 0[s] | loss 0.69\n",
      "| epoch 836 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 837 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 838 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 839 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 840 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 841 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 842 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 843 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 844 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 845 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 846 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 847 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 848 |  iter 1 / 2 | time 0[s] | loss 0.40\n",
      "| epoch 849 |  iter 1 / 2 | time 0[s] | loss 0.83\n",
      "| epoch 850 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 851 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 852 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 853 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 854 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 855 |  iter 1 / 2 | time 0[s] | loss 0.54\n",
      "| epoch 856 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 857 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 858 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 859 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 860 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 861 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 862 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 863 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 864 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 865 |  iter 1 / 2 | time 0[s] | loss 0.75\n",
      "| epoch 866 |  iter 1 / 2 | time 0[s] | loss 0.43\n",
      "| epoch 867 |  iter 1 / 2 | time 0[s] | loss 0.61\n",
      "| epoch 868 |  iter 1 / 2 | time 0[s] | loss 0.86\n",
      "| epoch 869 |  iter 1 / 2 | time 0[s] | loss 0.43\n",
      "| epoch 870 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 871 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 872 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 873 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 874 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 875 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 876 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 877 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 878 |  iter 1 / 2 | time 0[s] | loss 0.50\n",
      "| epoch 879 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 880 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 881 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 882 |  iter 1 / 2 | time 0[s] | loss 0.68\n",
      "| epoch 883 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 884 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 885 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 886 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 887 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 888 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 889 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 890 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 891 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 892 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 893 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 894 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 895 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 896 |  iter 1 / 2 | time 0[s] | loss 0.64\n",
      "| epoch 897 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 898 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 899 |  iter 1 / 2 | time 0[s] | loss 0.78\n",
      "| epoch 900 |  iter 1 / 2 | time 0[s] | loss 0.42\n",
      "| epoch 901 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 902 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 903 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 904 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 905 |  iter 1 / 2 | time 0[s] | loss 0.46\n",
      "| epoch 906 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 907 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 908 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 909 |  iter 1 / 2 | time 0[s] | loss 0.53\n",
      "| epoch 910 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 911 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 912 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 913 |  iter 1 / 2 | time 0[s] | loss 0.42\n",
      "| epoch 914 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 915 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 916 |  iter 1 / 2 | time 0[s] | loss 0.45\n",
      "| epoch 917 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 918 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 919 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 920 |  iter 1 / 2 | time 0[s] | loss 0.67\n",
      "| epoch 921 |  iter 1 / 2 | time 0[s] | loss 0.60\n",
      "| epoch 922 |  iter 1 / 2 | time 0[s] | loss 0.49\n",
      "| epoch 923 |  iter 1 / 2 | time 0[s] | loss 0.71\n",
      "| epoch 924 |  iter 1 / 2 | time 0[s] | loss 0.34\n",
      "| epoch 925 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 926 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 927 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 928 |  iter 1 / 2 | time 0[s] | loss 0.42\n",
      "| epoch 929 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 930 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 931 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 932 |  iter 1 / 2 | time 0[s] | loss 0.34\n",
      "| epoch 933 |  iter 1 / 2 | time 0[s] | loss 0.85\n",
      "| epoch 934 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 935 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 936 |  iter 1 / 2 | time 0[s] | loss 0.45\n",
      "| epoch 937 |  iter 1 / 2 | time 0[s] | loss 0.81\n",
      "| epoch 938 |  iter 1 / 2 | time 0[s] | loss 0.41\n",
      "| epoch 939 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 940 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 941 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 942 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 943 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 944 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 945 |  iter 1 / 2 | time 0[s] | loss 0.41\n",
      "| epoch 946 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 947 |  iter 1 / 2 | time 0[s] | loss 0.74\n",
      "| epoch 948 |  iter 1 / 2 | time 0[s] | loss 0.41\n",
      "| epoch 949 |  iter 1 / 2 | time 0[s] | loss 0.63\n",
      "| epoch 950 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 951 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 952 |  iter 1 / 2 | time 0[s] | loss 0.56\n",
      "| epoch 953 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 954 |  iter 1 / 2 | time 0[s] | loss 0.41\n",
      "| epoch 955 |  iter 1 / 2 | time 0[s] | loss 0.52\n",
      "| epoch 956 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 957 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 958 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 959 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 960 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 961 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 962 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 963 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 964 |  iter 1 / 2 | time 0[s] | loss 0.59\n",
      "| epoch 965 |  iter 1 / 2 | time 0[s] | loss 0.70\n",
      "| epoch 966 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 967 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 968 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 969 |  iter 1 / 2 | time 0[s] | loss 0.66\n",
      "| epoch 970 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 971 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 972 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 973 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 974 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 975 |  iter 1 / 2 | time 0[s] | loss 0.84\n",
      "| epoch 976 |  iter 1 / 2 | time 0[s] | loss 0.33\n",
      "| epoch 977 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 978 |  iter 1 / 2 | time 0[s] | loss 0.48\n",
      "| epoch 979 |  iter 1 / 2 | time 0[s] | loss 0.73\n",
      "| epoch 980 |  iter 1 / 2 | time 0[s] | loss 0.62\n",
      "| epoch 981 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 982 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 983 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 984 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 985 |  iter 1 / 2 | time 0[s] | loss 0.51\n",
      "| epoch 986 |  iter 1 / 2 | time 0[s] | loss 0.40\n",
      "| epoch 987 |  iter 1 / 2 | time 0[s] | loss 0.77\n",
      "| epoch 988 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 989 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 990 |  iter 1 / 2 | time 0[s] | loss 0.58\n",
      "| epoch 991 |  iter 1 / 2 | time 0[s] | loss 0.65\n",
      "| epoch 992 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 993 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 994 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 995 |  iter 1 / 2 | time 0[s] | loss 0.44\n",
      "| epoch 996 |  iter 1 / 2 | time 0[s] | loss 0.72\n",
      "| epoch 997 |  iter 1 / 2 | time 0[s] | loss 0.55\n",
      "| epoch 998 |  iter 1 / 2 | time 0[s] | loss 0.47\n",
      "| epoch 999 |  iter 1 / 2 | time 0[s] | loss 0.76\n",
      "| epoch 1000 |  iter 1 / 2 | time 0[s] | loss 0.51\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "you [ 1.1038777 -1.196002  -1.1143904 -1.3022822  1.1595826]\n",
      "say [-1.1599057   1.1478875   1.1967069   1.2020487  -0.29122028]\n",
      "goodbye [ 0.8487068  -0.7056579  -0.7870041  -0.546611    0.86524045]\n",
      "and [-0.9744709   0.93352073  0.8972963   1.0189264  -1.8906462 ]\n",
      "i [ 0.82690436 -0.70449865 -0.8036619  -0.5569182   0.87981224]\n",
      "hello [ 1.1011314 -1.2042462 -1.1150492 -1.2865607  1.181219 ]\n",
      ". [-1.0573411  1.0914458  1.1305699  1.0490285  1.830045 ]\n"
     ]
    }
   ],
   "source": [
    "# coding: utf-8\n",
    "import sys\n",
    "sys.path.append('..')  # 为导入父目录中的文件而进行的设置\n",
    "from common.trainer import Trainer\n",
    "from common.optimizer import Adam\n",
    "#from simple_cbow import SimpleCBOW\n",
    "from common.util import preprocess, create_contexts_target, convert_one_hot\n",
    "\n",
    "\n",
    "window_size = 1\n",
    "hidden_size = 5\n",
    "batch_size = 3\n",
    "max_epoch = 1000\n",
    "\n",
    "text = 'You say goodbye and I say hello.'\n",
    "corpus, word_to_id, id_to_word = preprocess(text)\n",
    "\n",
    "vocab_size = len(word_to_id)\n",
    "contexts, target = create_contexts_target(corpus, window_size)\n",
    "target = convert_one_hot(target, vocab_size)\n",
    "contexts = convert_one_hot(contexts, vocab_size)\n",
    "\n",
    "model = SimpleCBOW(vocab_size, hidden_size)\n",
    "optimizer = Adam()\n",
    "trainer = Trainer(model, optimizer)\n",
    "\n",
    "trainer.fit(contexts, target, max_epoch, batch_size)\n",
    "trainer.plot()\n",
    "\n",
    "word_vecs = model.word_vecs\n",
    "for word_id, word in id_to_word.items():\n",
    "    print(word, word_vecs[word_id])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# simple skip gram实现\n",
    "\n",
    "https://github.com/oreilly-japan/deep-learning-from-scratch-2/blob/master/ch03/simple_skip_gram.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# coding: utf-8\n",
    "import sys\n",
    "sys.path.append('..')\n",
    "import numpy as np\n",
    "from common.layers import MatMul, SoftmaxWithLoss\n",
    "\n",
    "\n",
    "class SimpleSkipGram:\n",
    "    def __init__(self, vocab_size, hidden_size):\n",
    "        V, H = vocab_size, hidden_size\n",
    "\n",
    "        #  权重初始化\n",
    "        W_in = 0.01 * np.random.randn(V, H).astype('f')\n",
    "        W_out = 0.01 * np.random.randn(H, V).astype('f')\n",
    "\n",
    "        # 生成层\n",
    "        self.in_layer = MatMul(W_in)\n",
    "        self.out_layer = MatMul(W_out)\n",
    "        self.loss_layer1 = SoftmaxWithLoss()\n",
    "        self.loss_layer2 = SoftmaxWithLoss()\n",
    "\n",
    "        # 将所有的权重和梯度汇总到列表\n",
    "        layers = [self.in_layer, self.out_layer]\n",
    "        self.params, self.grads = [], []\n",
    "        for layer in layers:\n",
    "            self.params += layer.params\n",
    "            self.grads += layer.grads\n",
    "\n",
    "        # 设置单词的离散表示成员变量\n",
    "        self.word_vecs = W_in\n",
    "\n",
    "    def forward(self, contexts, target):\n",
    "        h = self.in_layer.forward(target)\n",
    "        s = self.out_layer.forward(h)\n",
    "        l1 = self.loss_layer1.forward(s, contexts[:, 0])\n",
    "        l2 = self.loss_layer2.forward(s, contexts[:, 1])\n",
    "        loss = l1 + l2\n",
    "        return loss\n",
    "\n",
    "    def backward(self, dout=1):\n",
    "        dl1 = self.loss_layer1.backward(dout)\n",
    "        dl2 = self.loss_layer2.backward(dout)\n",
    "        ds = dl1 + dl2\n",
    "        dh = self.out_layer.backward(ds)\n",
    "        self.in_layer.backward(dh)\n",
    "        return None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 训练上述模型\n",
    "\n",
    "原作者未提供SimpleSkipGram的训练代码，这里参考前面的训练代码，编写了训练代码"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 1 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 2 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 3 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 4 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 5 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 6 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 7 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 8 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 9 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 10 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 11 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 12 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 13 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 14 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 15 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 16 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 17 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 18 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 19 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 20 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 21 |  iter 1 / 2 | time 0[s] | loss 3.89\n",
      "| epoch 22 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 23 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 24 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 25 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 26 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 27 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 28 |  iter 1 / 2 | time 0[s] | loss 3.88\n",
      "| epoch 29 |  iter 1 / 2 | time 0[s] | loss 3.87\n",
      "| epoch 30 |  iter 1 / 2 | time 0[s] | loss 3.87\n",
      "| epoch 31 |  iter 1 / 2 | time 0[s] | loss 3.87\n",
      "| epoch 32 |  iter 1 / 2 | time 0[s] | loss 3.87\n",
      "| epoch 33 |  iter 1 / 2 | time 0[s] | loss 3.87\n",
      "| epoch 34 |  iter 1 / 2 | time 0[s] | loss 3.86\n",
      "| epoch 35 |  iter 1 / 2 | time 0[s] | loss 3.86\n",
      "| epoch 36 |  iter 1 / 2 | time 0[s] | loss 3.86\n",
      "| epoch 37 |  iter 1 / 2 | time 0[s] | loss 3.86\n",
      "| epoch 38 |  iter 1 / 2 | time 0[s] | loss 3.85\n",
      "| epoch 39 |  iter 1 / 2 | time 0[s] | loss 3.85\n",
      "| epoch 40 |  iter 1 / 2 | time 0[s] | loss 3.85\n",
      "| epoch 41 |  iter 1 / 2 | time 0[s] | loss 3.85\n",
      "| epoch 42 |  iter 1 / 2 | time 0[s] | loss 3.84\n",
      "| epoch 43 |  iter 1 / 2 | time 0[s] | loss 3.84\n",
      "| epoch 44 |  iter 1 / 2 | time 0[s] | loss 3.84\n",
      "| epoch 45 |  iter 1 / 2 | time 0[s] | loss 3.83\n",
      "| epoch 46 |  iter 1 / 2 | time 0[s] | loss 3.83\n",
      "| epoch 47 |  iter 1 / 2 | time 0[s] | loss 3.83\n",
      "| epoch 48 |  iter 1 / 2 | time 0[s] | loss 3.83\n",
      "| epoch 49 |  iter 1 / 2 | time 0[s] | loss 3.81\n",
      "| epoch 50 |  iter 1 / 2 | time 0[s] | loss 3.82\n",
      "| epoch 51 |  iter 1 / 2 | time 0[s] | loss 3.81\n",
      "| epoch 52 |  iter 1 / 2 | time 0[s] | loss 3.81\n",
      "| epoch 53 |  iter 1 / 2 | time 0[s] | loss 3.79\n",
      "| epoch 54 |  iter 1 / 2 | time 0[s] | loss 3.80\n",
      "| epoch 55 |  iter 1 / 2 | time 0[s] | loss 3.80\n",
      "| epoch 56 |  iter 1 / 2 | time 0[s] | loss 3.79\n",
      "| epoch 57 |  iter 1 / 2 | time 0[s] | loss 3.78\n",
      "| epoch 58 |  iter 1 / 2 | time 0[s] | loss 3.79\n",
      "| epoch 59 |  iter 1 / 2 | time 0[s] | loss 3.76\n",
      "| epoch 60 |  iter 1 / 2 | time 0[s] | loss 3.77\n",
      "| epoch 61 |  iter 1 / 2 | time 0[s] | loss 3.77\n",
      "| epoch 62 |  iter 1 / 2 | time 0[s] | loss 3.76\n",
      "| epoch 63 |  iter 1 / 2 | time 0[s] | loss 3.75\n",
      "| epoch 64 |  iter 1 / 2 | time 0[s] | loss 3.76\n",
      "| epoch 65 |  iter 1 / 2 | time 0[s] | loss 3.74\n",
      "| epoch 66 |  iter 1 / 2 | time 0[s] | loss 3.73\n",
      "| epoch 67 |  iter 1 / 2 | time 0[s] | loss 3.74\n",
      "| epoch 68 |  iter 1 / 2 | time 0[s] | loss 3.73\n",
      "| epoch 69 |  iter 1 / 2 | time 0[s] | loss 3.72\n",
      "| epoch 70 |  iter 1 / 2 | time 0[s] | loss 3.71\n",
      "| epoch 71 |  iter 1 / 2 | time 0[s] | loss 3.70\n",
      "| epoch 72 |  iter 1 / 2 | time 0[s] | loss 3.71\n",
      "| epoch 73 |  iter 1 / 2 | time 0[s] | loss 3.70\n",
      "| epoch 74 |  iter 1 / 2 | time 0[s] | loss 3.69\n",
      "| epoch 75 |  iter 1 / 2 | time 0[s] | loss 3.68\n",
      "| epoch 76 |  iter 1 / 2 | time 0[s] | loss 3.68\n",
      "| epoch 77 |  iter 1 / 2 | time 0[s] | loss 3.68\n",
      "| epoch 78 |  iter 1 / 2 | time 0[s] | loss 3.64\n",
      "| epoch 79 |  iter 1 / 2 | time 0[s] | loss 3.66\n",
      "| epoch 80 |  iter 1 / 2 | time 0[s] | loss 3.65\n",
      "| epoch 81 |  iter 1 / 2 | time 0[s] | loss 3.67\n",
      "| epoch 82 |  iter 1 / 2 | time 0[s] | loss 3.63\n",
      "| epoch 83 |  iter 1 / 2 | time 0[s] | loss 3.63\n",
      "| epoch 84 |  iter 1 / 2 | time 0[s] | loss 3.65\n",
      "| epoch 85 |  iter 1 / 2 | time 0[s] | loss 3.62\n",
      "| epoch 86 |  iter 1 / 2 | time 0[s] | loss 3.58\n",
      "| epoch 87 |  iter 1 / 2 | time 0[s] | loss 3.59\n",
      "| epoch 88 |  iter 1 / 2 | time 0[s] | loss 3.64\n",
      "| epoch 89 |  iter 1 / 2 | time 0[s] | loss 3.58\n",
      "| epoch 90 |  iter 1 / 2 | time 0[s] | loss 3.60\n",
      "| epoch 91 |  iter 1 / 2 | time 0[s] | loss 3.53\n",
      "| epoch 92 |  iter 1 / 2 | time 0[s] | loss 3.60\n",
      "| epoch 93 |  iter 1 / 2 | time 0[s] | loss 3.57\n",
      "| epoch 94 |  iter 1 / 2 | time 0[s] | loss 3.51\n",
      "| epoch 95 |  iter 1 / 2 | time 0[s] | loss 3.57\n",
      "| epoch 96 |  iter 1 / 2 | time 0[s] | loss 3.54\n",
      "| epoch 97 |  iter 1 / 2 | time 0[s] | loss 3.56\n",
      "| epoch 98 |  iter 1 / 2 | time 0[s] | loss 3.49\n",
      "| epoch 99 |  iter 1 / 2 | time 0[s] | loss 3.52\n",
      "| epoch 100 |  iter 1 / 2 | time 0[s] | loss 3.47\n",
      "| epoch 101 |  iter 1 / 2 | time 0[s] | loss 3.56\n",
      "| epoch 102 |  iter 1 / 2 | time 0[s] | loss 3.45\n",
      "| epoch 103 |  iter 1 / 2 | time 0[s] | loss 3.50\n",
      "| epoch 104 |  iter 1 / 2 | time 0[s] | loss 3.50\n",
      "| epoch 105 |  iter 1 / 2 | time 0[s] | loss 3.46\n",
      "| epoch 106 |  iter 1 / 2 | time 0[s] | loss 3.48\n",
      "| epoch 107 |  iter 1 / 2 | time 0[s] | loss 3.47\n",
      "| epoch 108 |  iter 1 / 2 | time 0[s] | loss 3.41\n",
      "| epoch 109 |  iter 1 / 2 | time 0[s] | loss 3.44\n",
      "| epoch 110 |  iter 1 / 2 | time 0[s] | loss 3.47\n",
      "| epoch 111 |  iter 1 / 2 | time 0[s] | loss 3.42\n",
      "| epoch 112 |  iter 1 / 2 | time 0[s] | loss 3.38\n",
      "| epoch 113 |  iter 1 / 2 | time 0[s] | loss 3.42\n",
      "| epoch 114 |  iter 1 / 2 | time 0[s] | loss 3.37\n",
      "| epoch 115 |  iter 1 / 2 | time 0[s] | loss 3.41\n",
      "| epoch 116 |  iter 1 / 2 | time 0[s] | loss 3.42\n",
      "| epoch 117 |  iter 1 / 2 | time 0[s] | loss 3.31\n",
      "| epoch 118 |  iter 1 / 2 | time 0[s] | loss 3.38\n",
      "| epoch 119 |  iter 1 / 2 | time 0[s] | loss 3.36\n",
      "| epoch 120 |  iter 1 / 2 | time 0[s] | loss 3.30\n",
      "| epoch 121 |  iter 1 / 2 | time 0[s] | loss 3.42\n",
      "| epoch 122 |  iter 1 / 2 | time 0[s] | loss 3.31\n",
      "| epoch 123 |  iter 1 / 2 | time 0[s] | loss 3.32\n",
      "| epoch 124 |  iter 1 / 2 | time 0[s] | loss 3.27\n",
      "| epoch 125 |  iter 1 / 2 | time 0[s] | loss 3.34\n",
      "| epoch 126 |  iter 1 / 2 | time 0[s] | loss 3.27\n",
      "| epoch 127 |  iter 1 / 2 | time 0[s] | loss 3.37\n",
      "| epoch 128 |  iter 1 / 2 | time 0[s] | loss 3.25\n",
      "| epoch 129 |  iter 1 / 2 | time 0[s] | loss 3.31\n",
      "| epoch 130 |  iter 1 / 2 | time 0[s] | loss 3.28\n",
      "| epoch 131 |  iter 1 / 2 | time 0[s] | loss 3.23\n",
      "| epoch 132 |  iter 1 / 2 | time 0[s] | loss 3.23\n",
      "| epoch 133 |  iter 1 / 2 | time 0[s] | loss 3.28\n",
      "| epoch 134 |  iter 1 / 2 | time 0[s] | loss 3.18\n",
      "| epoch 135 |  iter 1 / 2 | time 0[s] | loss 3.25\n",
      "| epoch 136 |  iter 1 / 2 | time 0[s] | loss 3.21\n",
      "| epoch 137 |  iter 1 / 2 | time 0[s] | loss 3.21\n",
      "| epoch 138 |  iter 1 / 2 | time 0[s] | loss 3.17\n",
      "| epoch 139 |  iter 1 / 2 | time 0[s] | loss 3.15\n",
      "| epoch 140 |  iter 1 / 2 | time 0[s] | loss 3.21\n",
      "| epoch 141 |  iter 1 / 2 | time 0[s] | loss 3.18\n",
      "| epoch 142 |  iter 1 / 2 | time 0[s] | loss 3.20\n",
      "| epoch 143 |  iter 1 / 2 | time 0[s] | loss 3.17\n",
      "| epoch 144 |  iter 1 / 2 | time 0[s] | loss 3.12\n",
      "| epoch 145 |  iter 1 / 2 | time 0[s] | loss 3.21\n",
      "| epoch 146 |  iter 1 / 2 | time 0[s] | loss 3.03\n",
      "| epoch 147 |  iter 1 / 2 | time 0[s] | loss 3.19\n",
      "| epoch 148 |  iter 1 / 2 | time 0[s] | loss 3.13\n",
      "| epoch 149 |  iter 1 / 2 | time 0[s] | loss 3.06\n",
      "| epoch 150 |  iter 1 / 2 | time 0[s] | loss 3.08\n",
      "| epoch 151 |  iter 1 / 2 | time 0[s] | loss 3.16\n",
      "| epoch 152 |  iter 1 / 2 | time 0[s] | loss 3.08\n",
      "| epoch 153 |  iter 1 / 2 | time 0[s] | loss 3.12\n",
      "| epoch 154 |  iter 1 / 2 | time 0[s] | loss 3.01\n",
      "| epoch 155 |  iter 1 / 2 | time 0[s] | loss 3.17\n",
      "| epoch 156 |  iter 1 / 2 | time 0[s] | loss 2.97\n",
      "| epoch 157 |  iter 1 / 2 | time 0[s] | loss 3.05\n",
      "| epoch 158 |  iter 1 / 2 | time 0[s] | loss 3.13\n",
      "| epoch 159 |  iter 1 / 2 | time 0[s] | loss 2.98\n",
      "| epoch 160 |  iter 1 / 2 | time 0[s] | loss 2.99\n",
      "| epoch 161 |  iter 1 / 2 | time 0[s] | loss 3.03\n",
      "| epoch 162 |  iter 1 / 2 | time 0[s] | loss 3.09\n",
      "| epoch 163 |  iter 1 / 2 | time 0[s] | loss 2.89\n",
      "| epoch 164 |  iter 1 / 2 | time 0[s] | loss 2.95\n",
      "| epoch 165 |  iter 1 / 2 | time 0[s] | loss 3.10\n",
      "| epoch 166 |  iter 1 / 2 | time 0[s] | loss 2.99\n",
      "| epoch 167 |  iter 1 / 2 | time 0[s] | loss 2.89\n",
      "| epoch 168 |  iter 1 / 2 | time 0[s] | loss 3.00\n",
      "| epoch 169 |  iter 1 / 2 | time 0[s] | loss 2.91\n",
      "| epoch 170 |  iter 1 / 2 | time 0[s] | loss 2.91\n",
      "| epoch 171 |  iter 1 / 2 | time 0[s] | loss 2.94\n",
      "| epoch 172 |  iter 1 / 2 | time 0[s] | loss 3.02\n",
      "| epoch 173 |  iter 1 / 2 | time 0[s] | loss 2.88\n",
      "| epoch 174 |  iter 1 / 2 | time 0[s] | loss 2.93\n",
      "| epoch 175 |  iter 1 / 2 | time 0[s] | loss 2.96\n",
      "| epoch 176 |  iter 1 / 2 | time 0[s] | loss 2.96\n",
      "| epoch 177 |  iter 1 / 2 | time 0[s] | loss 2.80\n",
      "| epoch 178 |  iter 1 / 2 | time 0[s] | loss 2.99\n",
      "| epoch 179 |  iter 1 / 2 | time 0[s] | loss 2.89\n",
      "| epoch 180 |  iter 1 / 2 | time 0[s] | loss 2.93\n",
      "| epoch 181 |  iter 1 / 2 | time 0[s] | loss 2.72\n",
      "| epoch 182 |  iter 1 / 2 | time 0[s] | loss 2.80\n",
      "| epoch 183 |  iter 1 / 2 | time 0[s] | loss 2.96\n",
      "| epoch 184 |  iter 1 / 2 | time 0[s] | loss 2.80\n",
      "| epoch 185 |  iter 1 / 2 | time 0[s] | loss 2.90\n",
      "| epoch 186 |  iter 1 / 2 | time 0[s] | loss 2.74\n",
      "| epoch 187 |  iter 1 / 2 | time 0[s] | loss 2.93\n",
      "| epoch 188 |  iter 1 / 2 | time 0[s] | loss 2.71\n",
      "| epoch 189 |  iter 1 / 2 | time 0[s] | loss 2.93\n",
      "| epoch 190 |  iter 1 / 2 | time 0[s] | loss 2.75\n",
      "| epoch 191 |  iter 1 / 2 | time 0[s] | loss 2.96\n",
      "| epoch 192 |  iter 1 / 2 | time 0[s] | loss 2.69\n",
      "| epoch 193 |  iter 1 / 2 | time 0[s] | loss 2.72\n",
      "| epoch 194 |  iter 1 / 2 | time 0[s] | loss 2.90\n",
      "| epoch 195 |  iter 1 / 2 | time 0[s] | loss 2.79\n",
      "| epoch 196 |  iter 1 / 2 | time 0[s] | loss 2.70\n",
      "| epoch 197 |  iter 1 / 2 | time 0[s] | loss 2.66\n",
      "| epoch 198 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 199 |  iter 1 / 2 | time 0[s] | loss 2.87\n",
      "| epoch 200 |  iter 1 / 2 | time 0[s] | loss 2.74\n",
      "| epoch 201 |  iter 1 / 2 | time 0[s] | loss 2.75\n",
      "| epoch 202 |  iter 1 / 2 | time 0[s] | loss 2.73\n",
      "| epoch 203 |  iter 1 / 2 | time 0[s] | loss 2.73\n",
      "| epoch 204 |  iter 1 / 2 | time 0[s] | loss 2.78\n",
      "| epoch 205 |  iter 1 / 2 | time 0[s] | loss 2.59\n",
      "| epoch 206 |  iter 1 / 2 | time 0[s] | loss 2.77\n",
      "| epoch 207 |  iter 1 / 2 | time 0[s] | loss 2.65\n",
      "| epoch 208 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 209 |  iter 1 / 2 | time 0[s] | loss 2.64\n",
      "| epoch 210 |  iter 1 / 2 | time 0[s] | loss 2.82\n",
      "| epoch 211 |  iter 1 / 2 | time 0[s] | loss 2.55\n",
      "| epoch 212 |  iter 1 / 2 | time 0[s] | loss 2.61\n",
      "| epoch 213 |  iter 1 / 2 | time 0[s] | loss 2.80\n",
      "| epoch 214 |  iter 1 / 2 | time 0[s] | loss 2.66\n",
      "| epoch 215 |  iter 1 / 2 | time 0[s] | loss 2.73\n",
      "| epoch 216 |  iter 1 / 2 | time 0[s] | loss 2.45\n",
      "| epoch 217 |  iter 1 / 2 | time 0[s] | loss 2.79\n",
      "| epoch 218 |  iter 1 / 2 | time 0[s] | loss 2.63\n",
      "| epoch 219 |  iter 1 / 2 | time 0[s] | loss 2.64\n",
      "| epoch 220 |  iter 1 / 2 | time 0[s] | loss 2.56\n",
      "| epoch 221 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 222 |  iter 1 / 2 | time 0[s] | loss 2.41\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 223 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 224 |  iter 1 / 2 | time 0[s] | loss 2.54\n",
      "| epoch 225 |  iter 1 / 2 | time 0[s] | loss 2.67\n",
      "| epoch 226 |  iter 1 / 2 | time 0[s] | loss 2.54\n",
      "| epoch 227 |  iter 1 / 2 | time 0[s] | loss 2.66\n",
      "| epoch 228 |  iter 1 / 2 | time 0[s] | loss 2.59\n",
      "| epoch 229 |  iter 1 / 2 | time 0[s] | loss 2.59\n",
      "| epoch 230 |  iter 1 / 2 | time 0[s] | loss 2.72\n",
      "| epoch 231 |  iter 1 / 2 | time 0[s] | loss 2.29\n",
      "| epoch 232 |  iter 1 / 2 | time 0[s] | loss 2.71\n",
      "| epoch 233 |  iter 1 / 2 | time 0[s] | loss 2.64\n",
      "| epoch 234 |  iter 1 / 2 | time 0[s] | loss 2.49\n",
      "| epoch 235 |  iter 1 / 2 | time 0[s] | loss 2.62\n",
      "| epoch 236 |  iter 1 / 2 | time 0[s] | loss 2.49\n",
      "| epoch 237 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 238 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 239 |  iter 1 / 2 | time 0[s] | loss 2.47\n",
      "| epoch 240 |  iter 1 / 2 | time 0[s] | loss 2.45\n",
      "| epoch 241 |  iter 1 / 2 | time 0[s] | loss 2.61\n",
      "| epoch 242 |  iter 1 / 2 | time 0[s] | loss 2.52\n",
      "| epoch 243 |  iter 1 / 2 | time 0[s] | loss 2.36\n",
      "| epoch 244 |  iter 1 / 2 | time 0[s] | loss 2.76\n",
      "| epoch 245 |  iter 1 / 2 | time 0[s] | loss 2.42\n",
      "| epoch 246 |  iter 1 / 2 | time 0[s] | loss 2.51\n",
      "| epoch 247 |  iter 1 / 2 | time 0[s] | loss 2.43\n",
      "| epoch 248 |  iter 1 / 2 | time 0[s] | loss 2.72\n",
      "| epoch 249 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 250 |  iter 1 / 2 | time 0[s] | loss 2.49\n",
      "| epoch 251 |  iter 1 / 2 | time 0[s] | loss 2.48\n",
      "| epoch 252 |  iter 1 / 2 | time 0[s] | loss 2.41\n",
      "| epoch 253 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 254 |  iter 1 / 2 | time 0[s] | loss 2.54\n",
      "| epoch 255 |  iter 1 / 2 | time 0[s] | loss 2.65\n",
      "| epoch 256 |  iter 1 / 2 | time 0[s] | loss 2.46\n",
      "| epoch 257 |  iter 1 / 2 | time 0[s] | loss 2.36\n",
      "| epoch 258 |  iter 1 / 2 | time 0[s] | loss 2.46\n",
      "| epoch 259 |  iter 1 / 2 | time 0[s] | loss 2.45\n",
      "| epoch 260 |  iter 1 / 2 | time 0[s] | loss 2.62\n",
      "| epoch 261 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 262 |  iter 1 / 2 | time 0[s] | loss 2.51\n",
      "| epoch 263 |  iter 1 / 2 | time 0[s] | loss 2.27\n",
      "| epoch 264 |  iter 1 / 2 | time 0[s] | loss 2.60\n",
      "| epoch 265 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 266 |  iter 1 / 2 | time 0[s] | loss 2.69\n",
      "| epoch 267 |  iter 1 / 2 | time 0[s] | loss 2.08\n",
      "| epoch 268 |  iter 1 / 2 | time 0[s] | loss 2.48\n",
      "| epoch 269 |  iter 1 / 2 | time 0[s] | loss 2.45\n",
      "| epoch 270 |  iter 1 / 2 | time 0[s] | loss 2.30\n",
      "| epoch 271 |  iter 1 / 2 | time 0[s] | loss 2.68\n",
      "| epoch 272 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 273 |  iter 1 / 2 | time 0[s] | loss 2.36\n",
      "| epoch 274 |  iter 1 / 2 | time 0[s] | loss 2.43\n",
      "| epoch 275 |  iter 1 / 2 | time 0[s] | loss 2.46\n",
      "| epoch 276 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 277 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 278 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 279 |  iter 1 / 2 | time 0[s] | loss 2.27\n",
      "| epoch 280 |  iter 1 / 2 | time 0[s] | loss 2.48\n",
      "| epoch 281 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 282 |  iter 1 / 2 | time 0[s] | loss 2.48\n",
      "| epoch 283 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 284 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 285 |  iter 1 / 2 | time 0[s] | loss 2.54\n",
      "| epoch 286 |  iter 1 / 2 | time 0[s] | loss 2.35\n",
      "| epoch 287 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 288 |  iter 1 / 2 | time 0[s] | loss 2.53\n",
      "| epoch 289 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 290 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 291 |  iter 1 / 2 | time 0[s] | loss 2.40\n",
      "| epoch 292 |  iter 1 / 2 | time 0[s] | loss 2.45\n",
      "| epoch 293 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 294 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 295 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 296 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 297 |  iter 1 / 2 | time 0[s] | loss 2.32\n",
      "| epoch 298 |  iter 1 / 2 | time 0[s] | loss 2.62\n",
      "| epoch 299 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 300 |  iter 1 / 2 | time 0[s] | loss 2.31\n",
      "| epoch 301 |  iter 1 / 2 | time 0[s] | loss 2.31\n",
      "| epoch 302 |  iter 1 / 2 | time 0[s] | loss 2.49\n",
      "| epoch 303 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 304 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 305 |  iter 1 / 2 | time 0[s] | loss 2.42\n",
      "| epoch 306 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 307 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 308 |  iter 1 / 2 | time 0[s] | loss 2.53\n",
      "| epoch 309 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 310 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 311 |  iter 1 / 2 | time 0[s] | loss 2.41\n",
      "| epoch 312 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 313 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 314 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 315 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 316 |  iter 1 / 2 | time 0[s] | loss 2.40\n",
      "| epoch 317 |  iter 1 / 2 | time 0[s] | loss 2.27\n",
      "| epoch 318 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 319 |  iter 1 / 2 | time 0[s] | loss 2.31\n",
      "| epoch 320 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 321 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 322 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 323 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 324 |  iter 1 / 2 | time 0[s] | loss 2.30\n",
      "| epoch 325 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 326 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 327 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 328 |  iter 1 / 2 | time 0[s] | loss 2.57\n",
      "| epoch 329 |  iter 1 / 2 | time 0[s] | loss 2.09\n",
      "| epoch 330 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 331 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 332 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 333 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 334 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 335 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 336 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 337 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 338 |  iter 1 / 2 | time 0[s] | loss 2.32\n",
      "| epoch 339 |  iter 1 / 2 | time 0[s] | loss 2.30\n",
      "| epoch 340 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 341 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 342 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 343 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 344 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 345 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 346 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 347 |  iter 1 / 2 | time 0[s] | loss 2.40\n",
      "| epoch 348 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 349 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 350 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 351 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 352 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 353 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 354 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 355 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 356 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 357 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 358 |  iter 1 / 2 | time 0[s] | loss 2.38\n",
      "| epoch 359 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 360 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 361 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 362 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 363 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 364 |  iter 1 / 2 | time 0[s] | loss 2.31\n",
      "| epoch 365 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 366 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 367 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 368 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 369 |  iter 1 / 2 | time 0[s] | loss 2.31\n",
      "| epoch 370 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 371 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 372 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 373 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 374 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 375 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 376 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 377 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 378 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 379 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 380 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 381 |  iter 1 / 2 | time 0[s] | loss 2.50\n",
      "| epoch 382 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 383 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 384 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 385 |  iter 1 / 2 | time 0[s] | loss 2.08\n",
      "| epoch 386 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 387 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 388 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 389 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 390 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 391 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 392 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 393 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 394 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 395 |  iter 1 / 2 | time 0[s] | loss 2.44\n",
      "| epoch 396 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 397 |  iter 1 / 2 | time 0[s] | loss 2.07\n",
      "| epoch 398 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 399 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 400 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 401 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 402 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 403 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 404 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 405 |  iter 1 / 2 | time 0[s] | loss 1.76\n",
      "| epoch 406 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 407 |  iter 1 / 2 | time 0[s] | loss 2.06\n",
      "| epoch 408 |  iter 1 / 2 | time 0[s] | loss 2.35\n",
      "| epoch 409 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 410 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 411 |  iter 1 / 2 | time 0[s] | loss 2.30\n",
      "| epoch 412 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 413 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 414 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 415 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 416 |  iter 1 / 2 | time 0[s] | loss 2.29\n",
      "| epoch 417 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 418 |  iter 1 / 2 | time 0[s] | loss 2.09\n",
      "| epoch 419 |  iter 1 / 2 | time 0[s] | loss 2.50\n",
      "| epoch 420 |  iter 1 / 2 | time 0[s] | loss 1.87\n",
      "| epoch 421 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 422 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 423 |  iter 1 / 2 | time 0[s] | loss 2.43\n",
      "| epoch 424 |  iter 1 / 2 | time 0[s] | loss 1.86\n",
      "| epoch 425 |  iter 1 / 2 | time 0[s] | loss 2.07\n",
      "| epoch 426 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 427 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 428 |  iter 1 / 2 | time 0[s] | loss 2.07\n",
      "| epoch 429 |  iter 1 / 2 | time 0[s] | loss 2.07\n",
      "| epoch 430 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 431 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 432 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 433 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 434 |  iter 1 / 2 | time 0[s] | loss 2.06\n",
      "| epoch 435 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 436 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 437 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 438 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 439 |  iter 1 / 2 | time 0[s] | loss 2.27\n",
      "| epoch 440 |  iter 1 / 2 | time 0[s] | loss 2.09\n",
      "| epoch 441 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 442 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 443 |  iter 1 / 2 | time 0[s] | loss 2.06\n",
      "| epoch 444 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 445 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 446 |  iter 1 / 2 | time 0[s] | loss 2.02\n",
      "| epoch 447 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 448 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 449 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 450 |  iter 1 / 2 | time 0[s] | loss 1.83\n",
      "| epoch 451 |  iter 1 / 2 | time 0[s] | loss 2.26\n",
      "| epoch 452 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 453 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 454 |  iter 1 / 2 | time 0[s] | loss 1.83\n",
      "| epoch 455 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 456 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 457 |  iter 1 / 2 | time 0[s] | loss 2.07\n",
      "| epoch 458 |  iter 1 / 2 | time 0[s] | loss 1.82\n",
      "| epoch 459 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 460 |  iter 1 / 2 | time 0[s] | loss 2.47\n",
      "| epoch 461 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 462 |  iter 1 / 2 | time 0[s] | loss 2.28\n",
      "| epoch 463 |  iter 1 / 2 | time 0[s] | loss 1.85\n",
      "| epoch 464 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 465 |  iter 1 / 2 | time 0[s] | loss 2.25\n",
      "| epoch 466 |  iter 1 / 2 | time 0[s] | loss 2.06\n",
      "| epoch 467 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 468 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 469 |  iter 1 / 2 | time 0[s] | loss 2.03\n",
      "| epoch 470 |  iter 1 / 2 | time 0[s] | loss 2.03\n",
      "| epoch 471 |  iter 1 / 2 | time 0[s] | loss 1.84\n",
      "| epoch 472 |  iter 1 / 2 | time 0[s] | loss 1.99\n",
      "| epoch 473 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 474 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 475 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 476 |  iter 1 / 2 | time 0[s] | loss 2.27\n",
      "| epoch 477 |  iter 1 / 2 | time 0[s] | loss 1.80\n",
      "| epoch 478 |  iter 1 / 2 | time 0[s] | loss 2.02\n",
      "| epoch 479 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 480 |  iter 1 / 2 | time 0[s] | loss 1.99\n",
      "| epoch 481 |  iter 1 / 2 | time 0[s] | loss 2.02\n",
      "| epoch 482 |  iter 1 / 2 | time 0[s] | loss 1.99\n",
      "| epoch 483 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 484 |  iter 1 / 2 | time 0[s] | loss 2.02\n",
      "| epoch 485 |  iter 1 / 2 | time 0[s] | loss 2.05\n",
      "| epoch 486 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 487 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 488 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 489 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 490 |  iter 1 / 2 | time 0[s] | loss 1.82\n",
      "| epoch 491 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 492 |  iter 1 / 2 | time 0[s] | loss 2.23\n",
      "| epoch 493 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 494 |  iter 1 / 2 | time 0[s] | loss 2.04\n",
      "| epoch 495 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 496 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 497 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 498 |  iter 1 / 2 | time 0[s] | loss 2.17\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 499 |  iter 1 / 2 | time 0[s] | loss 2.06\n",
      "| epoch 500 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 501 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 502 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 503 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 504 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 505 |  iter 1 / 2 | time 0[s] | loss 1.81\n",
      "| epoch 506 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 507 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 508 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 509 |  iter 1 / 2 | time 0[s] | loss 2.02\n",
      "| epoch 510 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 511 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 512 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 513 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 514 |  iter 1 / 2 | time 0[s] | loss 2.22\n",
      "| epoch 515 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 516 |  iter 1 / 2 | time 0[s] | loss 1.58\n",
      "| epoch 517 |  iter 1 / 2 | time 0[s] | loss 2.41\n",
      "| epoch 518 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 519 |  iter 1 / 2 | time 0[s] | loss 2.24\n",
      "| epoch 520 |  iter 1 / 2 | time 0[s] | loss 1.57\n",
      "| epoch 521 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 522 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 523 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 524 |  iter 1 / 2 | time 0[s] | loss 2.01\n",
      "| epoch 525 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 526 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 527 |  iter 1 / 2 | time 0[s] | loss 1.79\n",
      "| epoch 528 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 529 |  iter 1 / 2 | time 0[s] | loss 1.76\n",
      "| epoch 530 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 531 |  iter 1 / 2 | time 0[s] | loss 1.59\n",
      "| epoch 532 |  iter 1 / 2 | time 0[s] | loss 2.21\n",
      "| epoch 533 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 534 |  iter 1 / 2 | time 0[s] | loss 1.99\n",
      "| epoch 535 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 536 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 537 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 538 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 539 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 540 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 541 |  iter 1 / 2 | time 0[s] | loss 1.78\n",
      "| epoch 542 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 543 |  iter 1 / 2 | time 0[s] | loss 1.76\n",
      "| epoch 544 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 545 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 546 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 547 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 548 |  iter 1 / 2 | time 0[s] | loss 2.00\n",
      "| epoch 549 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 550 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 551 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 552 |  iter 1 / 2 | time 0[s] | loss 2.20\n",
      "| epoch 553 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 554 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 555 |  iter 1 / 2 | time 0[s] | loss 1.77\n",
      "| epoch 556 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 557 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 558 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 559 |  iter 1 / 2 | time 0[s] | loss 1.57\n",
      "| epoch 560 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 561 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 562 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 563 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 564 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 565 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 566 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 567 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 568 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 569 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 570 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 571 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 572 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 573 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 574 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 575 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 576 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 577 |  iter 1 / 2 | time 0[s] | loss 2.19\n",
      "| epoch 578 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 579 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 580 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 581 |  iter 1 / 2 | time 0[s] | loss 1.76\n",
      "| epoch 582 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 583 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 584 |  iter 1 / 2 | time 0[s] | loss 1.73\n",
      "| epoch 585 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 586 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 587 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 588 |  iter 1 / 2 | time 0[s] | loss 1.73\n",
      "| epoch 589 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 590 |  iter 1 / 2 | time 0[s] | loss 1.73\n",
      "| epoch 591 |  iter 1 / 2 | time 0[s] | loss 2.39\n",
      "| epoch 592 |  iter 1 / 2 | time 0[s] | loss 1.53\n",
      "| epoch 593 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 594 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 595 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 596 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 597 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 598 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 599 |  iter 1 / 2 | time 0[s] | loss 1.54\n",
      "| epoch 600 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 601 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 602 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 603 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 604 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 605 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 606 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 607 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 608 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 609 |  iter 1 / 2 | time 0[s] | loss 1.97\n",
      "| epoch 610 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 611 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 612 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 613 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 614 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 615 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 616 |  iter 1 / 2 | time 0[s] | loss 1.98\n",
      "| epoch 617 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 618 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 619 |  iter 1 / 2 | time 0[s] | loss 1.75\n",
      "| epoch 620 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 621 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 622 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 623 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 624 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 625 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 626 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 627 |  iter 1 / 2 | time 0[s] | loss 1.73\n",
      "| epoch 628 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 629 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 630 |  iter 1 / 2 | time 0[s] | loss 1.96\n",
      "| epoch 631 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 632 |  iter 1 / 2 | time 0[s] | loss 2.18\n",
      "| epoch 633 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 634 |  iter 1 / 2 | time 0[s] | loss 1.74\n",
      "| epoch 635 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 636 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 637 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 638 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 639 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 640 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 641 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 642 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 643 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 644 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 645 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 646 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 647 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 648 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 649 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 650 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 651 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 652 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 653 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 654 |  iter 1 / 2 | time 0[s] | loss 2.17\n",
      "| epoch 655 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 656 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 657 |  iter 1 / 2 | time 0[s] | loss 1.95\n",
      "| epoch 658 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 659 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 660 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 661 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 662 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 663 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 664 |  iter 1 / 2 | time 0[s] | loss 1.72\n",
      "| epoch 665 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 666 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 667 |  iter 1 / 2 | time 0[s] | loss 1.50\n",
      "| epoch 668 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 669 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 670 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 671 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 672 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 673 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 674 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 675 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 676 |  iter 1 / 2 | time 0[s] | loss 2.15\n",
      "| epoch 677 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 678 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 679 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 680 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 681 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 682 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 683 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 684 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 685 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 686 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 687 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 688 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 689 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 690 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 691 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 692 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 693 |  iter 1 / 2 | time 0[s] | loss 1.94\n",
      "| epoch 694 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 695 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 696 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 697 |  iter 1 / 2 | time 0[s] | loss 2.16\n",
      "| epoch 698 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 699 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 700 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 701 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 702 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 703 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 704 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 705 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 706 |  iter 1 / 2 | time 0[s] | loss 2.36\n",
      "| epoch 707 |  iter 1 / 2 | time 0[s] | loss 1.49\n",
      "| epoch 708 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 709 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 710 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 711 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 712 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 713 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 714 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 715 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 716 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 717 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 718 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 719 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 720 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 721 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 722 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 723 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 724 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 725 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 726 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 727 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 728 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 729 |  iter 1 / 2 | time 0[s] | loss 1.93\n",
      "| epoch 730 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 731 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 732 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 733 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 734 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 735 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 736 |  iter 1 / 2 | time 0[s] | loss 1.71\n",
      "| epoch 737 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 738 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 739 |  iter 1 / 2 | time 0[s] | loss 2.37\n",
      "| epoch 740 |  iter 1 / 2 | time 0[s] | loss 1.47\n",
      "| epoch 741 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 742 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 743 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 744 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 745 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 746 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 747 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 748 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 749 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 750 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 751 |  iter 1 / 2 | time 0[s] | loss 2.14\n",
      "| epoch 752 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 753 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 754 |  iter 1 / 2 | time 0[s] | loss 2.35\n",
      "| epoch 755 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 756 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 757 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 758 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 759 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 760 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 761 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 762 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 763 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 764 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 765 |  iter 1 / 2 | time 0[s] | loss 1.46\n",
      "| epoch 766 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 767 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 768 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 769 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 770 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 771 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 772 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 773 |  iter 1 / 2 | time 0[s] | loss 1.70\n",
      "| epoch 774 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 775 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 776 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 777 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 778 |  iter 1 / 2 | time 0[s] | loss 1.69\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch 779 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 780 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 781 |  iter 1 / 2 | time 0[s] | loss 2.36\n",
      "| epoch 782 |  iter 1 / 2 | time 0[s] | loss 1.46\n",
      "| epoch 783 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 784 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 785 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 786 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 787 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 788 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 789 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 790 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 791 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 792 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 793 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 794 |  iter 1 / 2 | time 0[s] | loss 1.46\n",
      "| epoch 795 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 796 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 797 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 798 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 799 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 800 |  iter 1 / 2 | time 0[s] | loss 1.69\n",
      "| epoch 801 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 802 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 803 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 804 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 805 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 806 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 807 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 808 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 809 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 810 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 811 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 812 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 813 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 814 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 815 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 816 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 817 |  iter 1 / 2 | time 0[s] | loss 1.92\n",
      "| epoch 818 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 819 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 820 |  iter 1 / 2 | time 0[s] | loss 1.45\n",
      "| epoch 821 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 822 |  iter 1 / 2 | time 0[s] | loss 2.35\n",
      "| epoch 823 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 824 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 825 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 826 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 827 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 828 |  iter 1 / 2 | time 0[s] | loss 1.91\n",
      "| epoch 829 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 830 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 831 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 832 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 833 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 834 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 835 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 836 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 837 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 838 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 839 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 840 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 841 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 842 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 843 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 844 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 845 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 846 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 847 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 848 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 849 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 850 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 851 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 852 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 853 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 854 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 855 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 856 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 857 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 858 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 859 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 860 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 861 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 862 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 863 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 864 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 865 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 866 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 867 |  iter 1 / 2 | time 0[s] | loss 1.68\n",
      "| epoch 868 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 869 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 870 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 871 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 872 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 873 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 874 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 875 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 876 |  iter 1 / 2 | time 0[s] | loss 2.13\n",
      "| epoch 877 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 878 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 879 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 880 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 881 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 882 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 883 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 884 |  iter 1 / 2 | time 0[s] | loss 2.35\n",
      "| epoch 885 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 886 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 887 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 888 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 889 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 890 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 891 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 892 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 893 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 894 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 895 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 896 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 897 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 898 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 899 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 900 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 901 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 902 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 903 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 904 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 905 |  iter 1 / 2 | time 0[s] | loss 1.90\n",
      "| epoch 906 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 907 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 908 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 909 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 910 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 911 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 912 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 913 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 914 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 915 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 916 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 917 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 918 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 919 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 920 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 921 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 922 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 923 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 924 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 925 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 926 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 927 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 928 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 929 |  iter 1 / 2 | time 0[s] | loss 1.44\n",
      "| epoch 930 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 931 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 932 |  iter 1 / 2 | time 0[s] | loss 1.67\n",
      "| epoch 933 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 934 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 935 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 936 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 937 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 938 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 939 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 940 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 941 |  iter 1 / 2 | time 0[s] | loss 2.12\n",
      "| epoch 942 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 943 |  iter 1 / 2 | time 0[s] | loss 1.43\n",
      "| epoch 944 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 945 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 946 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 947 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 948 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 949 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 950 |  iter 1 / 2 | time 0[s] | loss 1.43\n",
      "| epoch 951 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 952 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 953 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 954 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 955 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 956 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 957 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 958 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 959 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 960 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 961 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 962 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 963 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 964 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 965 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 966 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 967 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 968 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 969 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 970 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 971 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 972 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 973 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 974 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 975 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 976 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 977 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 978 |  iter 1 / 2 | time 0[s] | loss 2.33\n",
      "| epoch 979 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 980 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 981 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 982 |  iter 1 / 2 | time 0[s] | loss 1.89\n",
      "| epoch 983 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 984 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 985 |  iter 1 / 2 | time 0[s] | loss 2.34\n",
      "| epoch 986 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 987 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 988 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 989 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 990 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 991 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 992 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 993 |  iter 1 / 2 | time 0[s] | loss 1.88\n",
      "| epoch 994 |  iter 1 / 2 | time 0[s] | loss 2.10\n",
      "| epoch 995 |  iter 1 / 2 | time 0[s] | loss 1.65\n",
      "| epoch 996 |  iter 1 / 2 | time 0[s] | loss 1.66\n",
      "| epoch 997 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 998 |  iter 1 / 2 | time 0[s] | loss 2.11\n",
      "| epoch 999 |  iter 1 / 2 | time 0[s] | loss 1.42\n",
      "| epoch 1000 |  iter 1 / 2 | time 0[s] | loss 1.89\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "you [ 0.00496249 -0.01331236 -0.01012976  0.00186353  0.00807451]\n",
      "say [ 0.16092646 -0.9065952  -0.9175136  -1.1712198   0.90137994]\n",
      "goodbye [ 1.0665348  0.8425567  0.8377366  0.9170987 -0.8304145]\n",
      "and [ 1.011805  -1.0424854 -1.0371493  1.3966451  1.0428121]\n",
      "i [ 1.0596083   0.84365463  0.82681113  0.922173   -0.8266081 ]\n",
      "hello [-1.2127174   0.99186623  0.9595063  -0.15925652 -0.96904427]\n",
      ". [ 0.01636839  0.00071803  0.00527886 -0.00629089 -0.00618663]\n"
     ]
    }
   ],
   "source": [
    "# coding: utf-8\n",
    "import sys\n",
    "sys.path.append('..')  # 为导入父目录中的文件而进行的设置\n",
    "from common.trainer import Trainer\n",
    "from common.optimizer import Adam\n",
    "from common.util import preprocess, create_contexts_target, convert_one_hot\n",
    "\n",
    "\n",
    "window_size = 1\n",
    "hidden_size = 5\n",
    "batch_size = 3\n",
    "max_epoch = 1000\n",
    "\n",
    "text = 'You say goodbye and I say hello.'\n",
    "corpus, word_to_id, id_to_word = preprocess(text)\n",
    "\n",
    "vocab_size = len(word_to_id)\n",
    "contexts, target = create_contexts_target(corpus, window_size)\n",
    "target = convert_one_hot(target, vocab_size)\n",
    "contexts = convert_one_hot(contexts, vocab_size)\n",
    "\n",
    "model = SimpleSkipGram(vocab_size, hidden_size)\n",
    "optimizer = Adam()\n",
    "trainer = Trainer(model, optimizer)\n",
    "\n",
    "trainer.fit(contexts, target, max_epoch, batch_size)\n",
    "trainer.plot()\n",
    "\n",
    "word_vecs = model.word_vecs\n",
    "for word_id, word in id_to_word.items():\n",
    "    print(word, word_vecs[word_id])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}