"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# 句子长度分布图\n",
"data['content'].apply(len).plot(kind='box')\n",
"plt.title('Data')"
]
},
{
"cell_type": "markdown",
"id": "f29ad2fb",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"句子长度分布数据：作文长度一般是 1015 词左右"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "23bf33fa-9948-43a7-95a6-ed212bd5ffc2",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:17:01.337236Z",
"start_time": "2023-08-08T12:17:01.320931Z"
},
"execution": {
"iopub.execute_input": "2023-08-02T06:39:46.002602Z",
"iopub.status.busy": "2023-08-02T06:39:46.002369Z",
"iopub.status.idle": "2023-08-02T06:39:46.016465Z",
"shell.execute_reply": "2023-08-02T06:39:46.015707Z",
"shell.execute_reply.started": "2023-08-02T06:39:46.002582Z"
},
"scrolled": true,
"slideshow": {
"slide_type": "-"
}
},
"outputs": [
{
"data": {
"text/plain": [
"count 14000.000000\n",
"mean 1014.411857\n",
"std 16.114179\n",
"min 2.000000\n",
"25% 1015.000000\n",
"50% 1015.000000\n",
"75% 1015.000000\n",
"max 1015.000000\n",
"Name: content, dtype: float64"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# 句子长度分布数据\n",
"data[\"content\"].apply(len).describe()"
]
},
{
"cell_type": "markdown",
"id": "112e945c",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"标签分布图：不均衡，0 类远多于 1 类"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "e476683d-6a6c-4b6c-a5d3-56b4f718db80",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:18:38.413625Z",
"start_time": "2023-08-08T12:18:38.320922Z"
},
"execution": {
"iopub.execute_input": "2023-08-02T06:39:46.017620Z",
"iopub.status.busy": "2023-08-02T06:39:46.017400Z",
"iopub.status.idle": "2023-08-02T06:39:46.111803Z",
"shell.execute_reply": "2023-08-02T06:39:46.111010Z",
"shell.execute_reply.started": "2023-08-02T06:39:46.017601Z"
},
"scrolled": true,
"slideshow": {
"slide_type": "-"
}
},
"outputs": [
{
"data": {
"text/plain": [
"Text(1, 1, 'label\\n0 11836\\n1 2164\\nName: count, dtype: int64')"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
"
"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# 标签分布图\n",
"data[\"label\"].value_counts().plot(kind=\"pie\").text(1, 1, f\"{data['label'].value_counts()}\")"
]
},
{
"cell_type": "markdown",
"id": "bb54bb06",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# 模型定义与训练"
]
},
{
"cell_type": "markdown",
"id": "344ad03a",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"数据集格式转换"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "57ba0b71-0d9a-4375-9a04-fe5baad48abc",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:31:21.013194Z",
"start_time": "2023-08-08T12:31:20.986025Z"
},
"execution": {
"iopub.execute_input": "2023-08-02T06:40:23.367224Z",
"iopub.status.busy": "2023-08-02T06:40:23.365916Z",
"iopub.status.idle": "2023-08-02T06:40:23.400959Z",
"shell.execute_reply": "2023-08-02T06:40:23.399898Z",
"shell.execute_reply.started": "2023-08-02T06:40:23.367180Z"
},
"run_control": {
"marked": false
},
"scrolled": true,
"slideshow": {
"slide_type": "-"
},
"tags": []
},
"outputs": [],
"source": [
"# 按照 10% 的比例划分训练集与验证集\n",
"train_data, valid_data = train_test_split(data[:10], test_size=0.1)\n",
"\n",
"# 下面就是一堆操作，把数据变成数据加载器可以识别的格式，自定义数据集类也是同样的效果\n",
"train_dict = train_data.to_dict(orient='records')\n",
"valid_dict = valid_data.to_dict(orient='records')\n",
"train_ds = MapDataset(train_dict)\n",
"valid_ds = MapDataset(valid_dict)\n",
"\n",
"# 将整体数据拆分为 <总样本数 / 2> 份，放入数据加载器，就是一次性会有 2 份数据同时并行计算，份数越多，并行越多，显存占用越大，需要根据需求来选择\n",
"train_loader = DataLoader(train_ds, batch_size=2, shuffle=True) # 训练数据可以随机打乱，让模型更好地学习，减轻学习到无关特征的问题\n",
"valid_loader = DataLoader(valid_ds, batch_size=2) # 这里用的是 1650TI 4G，如果是显存更小的卡，需要调小一点，不然会炸显存"
]
},
{
"cell_type": "markdown",
"id": "07efaf4c",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"模型定义"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "1f3f209b-5aa7-4aa2-9e45-591d01523b28",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:42:48.180177Z",
"start_time": "2023-08-08T12:42:41.108951Z"
},
"scrolled": true,
"slideshow": {
"slide_type": "-"
},
"tags": []
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[32m[2023-08-08 20:42:41,111] [ INFO]\u001b[0m - We are using to load 'ernie-3.0-mini-zh'.\u001b[0m\n",
"\u001b[32m[2023-08-08 20:42:41,114] [ INFO]\u001b[0m - Model config ErnieConfig {\n",
" \"attention_probs_dropout_prob\": 0.1,\n",
" \"enable_recompute\": false,\n",
" \"fuse\": false,\n",
" \"hidden_act\": \"gelu\",\n",
" \"hidden_dropout_prob\": 0.1,\n",
" \"hidden_size\": 384,\n",
" \"initializer_range\": 0.02,\n",
" \"intermediate_size\": 1536,\n",
" \"layer_norm_eps\": 1e-12,\n",
" \"max_position_embeddings\": 2048,\n",
" \"model_type\": \"ernie\",\n",
" \"num_attention_heads\": 12,\n",
" \"num_hidden_layers\": 6,\n",
" \"pad_token_id\": 0,\n",
" \"paddlenlp_version\": null,\n",
" \"pool_act\": \"tanh\",\n",
" \"task_id\": 0,\n",
" \"task_type_vocab_size\": 16,\n",
" \"type_vocab_size\": 4,\n",
" \"use_task_id\": true,\n",
" \"vocab_size\": 40000\n",
"}\n",
"\u001b[0m\n",
"\u001b[33m[2023-08-08 20:42:48,121] [ WARNING]\u001b[0m - Some weights of the model checkpoint at ernie-3.0-mini-zh were not used when initializing ErnieForSequenceClassification: ['ernie.encoder.layers.6.self_attn.k_proj.bias', 'ernie.encoder.layers.6.norm1.bias', 'ernie.encoder.layers.6.norm2.bias', 'ernie.encoder.layers.6.linear1.weight', 'ernie.encoder.layers.6.self_attn.out_proj.weight', 'ernie.encoder.layers.6.self_attn.v_proj.bias', 'ernie.encoder.layers.6.linear2.bias', 'ernie.encoder.layers.6.linear1.bias', 'ernie.encoder.layers.6.self_attn.v_proj.weight', 'ernie.encoder.layers.6.self_attn.k_proj.weight', 'ernie.encoder.layers.6.norm2.weight', 'ernie.encoder.layers.6.self_attn.q_proj.weight', 'ernie.encoder.layers.6.norm1.weight', 'ernie.encoder.layers.6.self_attn.q_proj.bias', 'ernie.encoder.layers.6.self_attn.out_proj.bias', 'ernie.encoder.layers.6.linear2.weight']\n",
"- This IS expected if you are initializing ErnieForSequenceClassification from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
"- This IS NOT expected if you are initializing ErnieForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\u001b[0m\n",
"\u001b[33m[2023-08-08 20:42:48,122] [ WARNING]\u001b[0m - Some weights of ErnieForSequenceClassification were not initialized from the model checkpoint at ernie-3.0-mini-zh and are newly initialized: ['classifier.bias', 'classifier.weight', 'ernie.pooler.dense.bias', 'ernie.pooler.dense.weight']\n",
"You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\u001b[0m\n",
"\u001b[32m[2023-08-08 20:42:48,125] [ INFO]\u001b[0m - We are using to load 'ernie-3.0-mini-zh'.\u001b[0m\n",
"\u001b[32m[2023-08-08 20:42:48,127] [ INFO]\u001b[0m - Already cached C:\\Users\\Epsilon_Luoo\\.paddlenlp\\models\\ernie-3.0-mini-zh\\ernie_3.0_mini_zh_vocab.txt\u001b[0m\n",
"\u001b[32m[2023-08-08 20:42:48,155] [ INFO]\u001b[0m - tokenizer config file saved in C:\\Users\\Epsilon_Luoo\\.paddlenlp\\models\\ernie-3.0-mini-zh\\tokenizer_config.json\u001b[0m\n",
"\u001b[32m[2023-08-08 20:42:48,161] [ INFO]\u001b[0m - Special tokens file saved in C:\\Users\\Epsilon_Luoo\\.paddlenlp\\models\\ernie-3.0-mini-zh\\special_tokens_map.json\u001b[0m\n"
]
}
],
"source": [
"# 载入模型与分词器\n",
"\n",
"# 使用 ernie-3.0-mini-zh 序列分类模型，并将分类类别数设置为 2\n",
"model = AutoModelForSequenceClassification.from_pretrained(\"ernie-3.0-mini-zh\", num_classes=2)\n",
"# 使用 ernie-3.0-mini-zh 分词器\n",
"tokenizer = AutoTokenizer.from_pretrained(\"ernie-3.0-mini-zh\")\n",
"\n",
"# 定义 AdamW 优化器，学习率为 0.000001\n",
"optimizer = paddle.optimizer.AdamW(1e-5, parameters=model.parameters())\n",
"\n",
"# 定义损失函数为交叉熵函数，计算每个 mini batch 的均值\n",
"loss_fn = paddle.nn.loss.CrossEntropyLoss(reduction='mean')"
]
},
{
"cell_type": "markdown",
"id": "602683bf",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"模型训练"
]
},
{
"cell_type": "code",
"execution_count": 14,
"id": "5034a124-1d12-45bc-9e71-ed079bb115c0",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:43:02.220455Z",
"start_time": "2023-08-08T12:43:00.944664Z"
},
"execution": {
"iopub.execute_input": "2023-08-02T06:40:25.081247Z",
"iopub.status.busy": "2023-08-02T06:40:25.080031Z",
"iopub.status.idle": "2023-08-02T07:15:16.566927Z",
"shell.execute_reply": "2023-08-02T07:15:16.566014Z",
"shell.execute_reply.started": "2023-08-02T06:40:25.081213Z"
},
"run_control": {
"marked": false
},
"scrolled": true,
"tags": []
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:00<00:00, 10.36it/s]\n",
"100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 52.45it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 0, Val loss 0.319446, Val Accuracy 1.000000\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:00<00:00, 19.81it/s]\n",
"100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 57.42it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1, Val loss 0.265423, Val Accuracy 1.000000\n"
]
}
],
"source": [
"for epoch in range(2): # 训练 2 轮\n",
" # 训练过程\n",
" model.train() # 切换模型为训练模式\n",
" for batch_x in tqdm(train_loader): # 每次从数据加载器读入一批(batch)数据\n",
" X = tokenizer(batch_x[\"content\"], max_length=1015, padding=True) # 将数据转换为模块可处理的数据形式\n",
" input_ids = paddle.to_tensor(X['input_ids'], dtype=\"int32\") # 将 input_ids 变为张量，方便并行计算\n",
" token_type_ids = paddle.to_tensor(X['token_type_ids'], dtype=\"int32\") # 将 token_type_ids 变为张量\n",
" pred = model(input_ids, token_type_ids) # 将数据读入模型，并得到计算后的结果\n",
" loss = loss_fn(pred, paddle.to_tensor(batch_x[\"label\"], dtype=\"int32\")) # 对比预测结果与真实结果，计算损失函数的值\n",
" loss.backward() # 反向传播，计算梯度\n",
" optimizer.step() # 优化器根据梯度与学习率调整模型参数\n",
" optimizer.clear_gradients() # 清空梯度，避免下次梯度计算时累加\n",
"\n",
" # 验证过程\n",
" model.eval() # 切换模型为验证模式\n",
" val_loss = [] # 验证集数据的损失函数合集\n",
" with paddle.no_grad(): # 在模型验证时，只做前向计算，因此不需要保存梯度信息\n",
" for batch_x in tqdm(valid_loader): # 下面的操作与训练过程相同\n",
" X = tokenizer(batch_x[\"content\"], max_length=1015, padding=True)\n",
" input_ids = paddle.to_tensor(X['input_ids'], dtype=\"int32\")\n",
" token_type_ids = paddle.to_tensor(X['token_type_ids'], dtype=\"int32\")\n",
" pred = model(input_ids, token_type_ids)\n",
" loss = loss_fn(pred, paddle.to_tensor(batch_x[\"label\"], dtype=\"int32\"))\n",
" val_loss.append(loss.item()) # 将计算出的损失函数值存入合集\n",
" \n",
" # 打印本轮训练的验证集损失函数值，与预测正确率\n",
" print('Epoch {0}, Val loss {1:3f}, Val Accuracy {2:3f}'.format(\n",
" epoch,\n",
" np.mean(val_loss), \n",
" (pred.argmax(1) == batch_x[\"label\"]).astype('float').mean().item()\n",
"))\n",
"# 保存模型参数\n",
"paddle.save(model.state_dict(), \"work/model.pdparams\")\n",
"# 保存优化器参数\n",
"paddle.save(optimizer.state_dict(), \"work/optimizer.pdopt\")"
]
},
{
"cell_type": "markdown",
"id": "4171baba",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# 模型载入"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7c2df496-3c9a-42b2-851d-e793c9a3479d",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T08:23:00.580208Z",
"start_time": "2023-08-08T08:22:59.550174Z"
},
"execution": {
"iopub.execute_input": "2023-08-03T02:25:23.013252Z",
"iopub.status.busy": "2023-08-03T02:25:23.012568Z",
"iopub.status.idle": "2023-08-03T02:25:23.596904Z",
"shell.execute_reply": "2023-08-03T02:25:23.595810Z",
"shell.execute_reply.started": "2023-08-03T02:25:23.013220Z"
},
"scrolled": false,
"tags": []
},
"outputs": [],
"source": [
"# 如果你拿到了模型参数（在 AIStudio 中提供），你可以运行这行代码，如果直接运行模型，则没有必要运行\n",
"\n",
"# 载入模型参数、优化器参数的最后一个epoch保存的检查点\n",
"layer_state_dict = paddle.load(\"model.pdparams\")\n",
"opt_state_dict = paddle.load(\"optimizer.pdopt\")\n",
"\n",
"# 将加载后的参数与模型关联起来\n",
"model.set_state_dict(layer_state_dict)\n",
"optimizer.set_state_dict(opt_state_dict)"
]
},
{
"cell_type": "markdown",
"id": "56fd0304",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# 推理过程"
]
},
{
"cell_type": "code",
"execution_count": 17,
"id": "dae1075c-d68b-4b63-8e46-9794bfd4d692",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:47:50.622747Z",
"start_time": "2023-08-08T12:47:50.612233Z"
},
"execution": {
"iopub.execute_input": "2023-08-03T02:25:29.238158Z",
"iopub.status.busy": "2023-08-03T02:25:29.237602Z",
"iopub.status.idle": "2023-08-03T02:25:29.245148Z",
"shell.execute_reply": "2023-08-03T02:25:29.244150Z",
"shell.execute_reply.started": "2023-08-03T02:25:29.238127Z"
},
"scrolled": true,
"tags": []
},
"outputs": [],
"source": [
"# 自定义推理函数\n",
"def infer(string: str) -> int:\n",
" \"\"\"将文本传入模型并返回预测结果\n",
" \n",
" 输入：\n",
" - string: str\n",
" 待预测的文本内容\n",
" \n",
" 输出:\n",
" - result: int\n",
" 模型的预测结果\n",
" \"\"\"\n",
" \n",
" X = tokenizer([string], max_length=1015, padding=True)\n",
" input_ids = paddle.to_tensor(X['input_ids'], dtype=\"int32\")\n",
" token_type_ids = paddle.to_tensor(X['token_type_ids'], dtype=\"int32\")\n",
" pred = model(input_ids, token_type_ids)\n",
" result = pred.argmax(1).item() # 获取预测概率最大的那个类别\n",
" return result"
]
},
{
"cell_type": "markdown",
"id": "18b1d2cc",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"提交文件生成"
]
},
{
"cell_type": "code",
"execution_count": 18,
"id": "836e49c2-f980-4626-8a43-ab3bdfa65a69",
"metadata": {
"ExecuteTime": {
"end_time": "2023-08-08T12:47:54.567036Z",
"start_time": "2023-08-08T12:47:52.785648Z"
},
"execution": {
"iopub.execute_input": "2023-08-02T07:17:04.890315Z",
"iopub.status.busy": "2023-08-02T07:17:04.889690Z",
"iopub.status.idle": "2023-08-02T07:19:00.263819Z",
"shell.execute_reply": "2023-08-02T07:19:00.262948Z",
"shell.execute_reply.started": "2023-08-02T07:17:04.890273Z"
},
"run_control": {
"marked": false
},
"scrolled": true,
"tags": []
},
"outputs": [
{
"data": {
"text/html": [
"![](\"datawhale_wechat_qrcode.jpeg\")
"
]
}
],
"metadata": {
"celltoolbar": "Slideshow",
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"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": "3.10.12"
},
"papermill": {
"default_parameters": {},
"duration": 17022.610815,
"end_time": "2023-03-24T09:19:21.540649",
"environment_variables": {},
"exception": null,
"input_path": "/home/aistudio/main.ipynb",
"output_path": "/home/aistudio/.main.ipynb",
"parameters": {},
"start_time": "2023-03-24T04:35:38.929834",
"version": "2.3.4"
},
"rise": {
"overlay": "
"
},
"varInspector": {
"cols": {
"lenName": 16,
"lenType": 16,
"lenVar": 40
},
"kernels_config": {
"python": {
"delete_cmd_postfix": "",
"delete_cmd_prefix": "del ",
"library": "var_list.py",
"varRefreshCmd": "print(var_dic_list())"
},
"r": {
"delete_cmd_postfix": ") ",
"delete_cmd_prefix": "rm(",
"library": "var_list.r",
"varRefreshCmd": "cat(var_dic_list()) "
}
},
"types_to_exclude": [
"module",
"function",
"builtin_function_or_method",
"instance",
"_Feature"
],
"window_display": false
}
},
"nbformat": 4,
"nbformat_minor": 5
}
\n",
"\n",
"\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" name content label\n",
"0 14001 [3125 2196 286 123 1539 1759 266 3549 649 ... 0.0\n",
"1 14002 [1109 2113 3122 213 3125 1294 5212 2338 2233 ... 0.0\n",
"2 14003 [ 236 3125 139 3037 5212 4294 1600 4550 3169 ... 0.0\n",
"3 14004 [ 13 13 13 0 0 0 245 1472 3125 ... 0.0\n",
"4 14005 [2113 2444 139 1109 4648 4626 181 3635 1145 ... 0.0\n",
"5 14006 [ 664 139 220 1759 248 2188 664 4544 3125 ... 0.0\n",
"6 14007 [1472 2214 5212 1759 3125 1294 199 675 3037 ... 0.0\n",
"7 14008 [ 13 0 0 0 216 2176 3125 526 1100 ... 0.0\n",
"8 14009 [5212 1339 3747 4242 1759 266 2101 3122 1115 ... 0.0\n",
"9 14010 [3125 2278 983 4982 2138 284 3635 5212 2113 ... 0.0"
]
},
"execution_count": 18,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"test_data[\"label\"] = test_data[\"content\"][:100].apply(infer) # 将测试集的每个文本送入模型返回结果\n",
"submit = test_data.drop(columns=[\"content\"]) # 生成提交数据（就是把带结果的测试集丢掉内容，复制一份）\n",
"submit.to_csv(\"submit.csv\", index=False) # 保存 CSV 文件\n",
"test_data.head(10)"
]
},
{
"cell_type": "markdown",
"id": "dffcae16",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# Thank You ;-)\n",
"Datawhale 聪明办法学 Python 教学团队出品\n",
"\n",
"## 关注我们\n",
"Datawhale 是一个专注 AI 领域的开源组织，以“for the learner，和学习者一起成长”为愿景，构建对学习者最有价值的开源学习社区。关注我们，一起学习成长。\n",
"| \n", " | name | \n", "content | \n", "label | \n", "
|---|---|---|---|
| 0 | \n", "14001 | \n", "[3125 2196 286 123 1539 1759 266 3549 649 ... | \n", "0.0 | \n", "
| 1 | \n", "14002 | \n", "[1109 2113 3122 213 3125 1294 5212 2338 2233 ... | \n", "0.0 | \n", "
| 2 | \n", "14003 | \n", "[ 236 3125 139 3037 5212 4294 1600 4550 3169 ... | \n", "0.0 | \n", "
| 3 | \n", "14004 | \n", "[ 13 13 13 0 0 0 245 1472 3125 ... | \n", "0.0 | \n", "
| 4 | \n", "14005 | \n", "[2113 2444 139 1109 4648 4626 181 3635 1145 ... | \n", "0.0 | \n", "
| 5 | \n", "14006 | \n", "[ 664 139 220 1759 248 2188 664 4544 3125 ... | \n", "0.0 | \n", "
| 6 | \n", "14007 | \n", "[1472 2214 5212 1759 3125 1294 199 675 3037 ... | \n", "0.0 | \n", "
| 7 | \n", "14008 | \n", "[ 13 0 0 0 216 2176 3125 526 1100 ... | \n", "0.0 | \n", "
| 8 | \n", "14009 | \n", "[5212 1339 3747 4242 1759 266 2101 3122 1115 ... | \n", "0.0 | \n", "
| 9 | \n", "14010 | \n", "[3125 2278 983 4982 2138 284 3635 5212 2113 ... | \n", "0.0 | \n", "