Upload 8 files
Browse files- adapter_config.json +22 -0
- adapter_model.bin +3 -0
- added_tokens.json +3 -0
- finetune.ipynb +1130 -0
- special_tokens_map.json +24 -0
- tokenizer.json +0 -0
- tokenizer.model +3 -0
- tokenizer_config.json +34 -0
adapter_config.json
ADDED
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"base_model_name_or_path": "/workspace/WizardLM-7b",
|
3 |
+
"bias": "all",
|
4 |
+
"fan_in_fan_out": false,
|
5 |
+
"inference_mode": true,
|
6 |
+
"init_lora_weights": true,
|
7 |
+
"lora_alpha": 64,
|
8 |
+
"lora_dropout": 0.1,
|
9 |
+
"modules_to_save": null,
|
10 |
+
"peft_type": "LORA",
|
11 |
+
"r": 8,
|
12 |
+
"target_modules": [
|
13 |
+
"q_proj",
|
14 |
+
"k_proj",
|
15 |
+
"v_proj",
|
16 |
+
"out_proj",
|
17 |
+
"fc_in",
|
18 |
+
"fc_out",
|
19 |
+
"wte"
|
20 |
+
],
|
21 |
+
"task_type": "CAUSAL_LM"
|
22 |
+
}
|
adapter_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d7e509a37773ddff40c0df2df2979f2da94c3e92374b6ef8903962172a871397
|
3 |
+
size 25234701
|
added_tokens.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"[PAD]": 32000
|
3 |
+
}
|
finetune.ipynb
ADDED
@@ -0,0 +1,1130 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": 1,
|
6 |
+
"id": "3139c31d-2ca2-472e-8e29-c404032e4a55",
|
7 |
+
"metadata": {},
|
8 |
+
"outputs": [],
|
9 |
+
"source": [
|
10 |
+
"from pathlib import Path\n",
|
11 |
+
"\n",
|
12 |
+
"#!git clone https://github.com/IlyaGusev/rulm.git\n",
|
13 |
+
"#!git clone https://github.com/ggerganov/llama.cpp.git\n",
|
14 |
+
"\n",
|
15 |
+
"self_instruct_dir = Path('rulm/self_instruct').resolve()\n",
|
16 |
+
"\n",
|
17 |
+
"#!cd {self_instruct_dir} && pip install -r ../requirements.txt"
|
18 |
+
]
|
19 |
+
},
|
20 |
+
{
|
21 |
+
"cell_type": "code",
|
22 |
+
"execution_count": 2,
|
23 |
+
"id": "6f04525b-5567-4335-98b2-7bb60d4be6ca",
|
24 |
+
"metadata": {},
|
25 |
+
"outputs": [],
|
26 |
+
"source": [
|
27 |
+
"from huggingface_hub import snapshot_download\n",
|
28 |
+
"content_dir = Path('.').resolve()\n",
|
29 |
+
"\n",
|
30 |
+
"model_dir = content_dir / \"WizardLM-7b\"\n",
|
31 |
+
"base_model = \"ehartford/WizardLM-7B-V1.0-Uncensored\" #@param {type:\"string\"}\n",
|
32 |
+
"#snapshot_download(repo_id=base_model, local_dir=model_dir, ignore_patterns=[\"LICENSE\", \"README.md\", \".gitattributes\"])\n"
|
33 |
+
]
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"cell_type": "code",
|
37 |
+
"execution_count": 3,
|
38 |
+
"id": "5321ea58-116a-4501-b6c3-3521ac5e6e7f",
|
39 |
+
"metadata": {},
|
40 |
+
"outputs": [
|
41 |
+
{
|
42 |
+
"name": "stderr",
|
43 |
+
"output_type": "stream",
|
44 |
+
"text": [
|
45 |
+
"Overriding torch_dtype=None with `torch_dtype=torch.float16` due to requirements of `bitsandbytes` to enable model loading in mixed int8. Either pass torch_dtype=torch.float16 or don't pass this argument at all to remove this warning.\n"
|
46 |
+
]
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"name": "stdout",
|
50 |
+
"output_type": "stream",
|
51 |
+
"text": [
|
52 |
+
"\n",
|
53 |
+
"===================================BUG REPORT===================================\n",
|
54 |
+
"Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues\n",
|
55 |
+
"================================================================================\n",
|
56 |
+
"CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...\n",
|
57 |
+
"CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so\n",
|
58 |
+
"CUDA SETUP: Highest compute capability among GPUs detected: 8.0\n",
|
59 |
+
"CUDA SETUP: Detected CUDA version 117\n",
|
60 |
+
"CUDA SETUP: Loading binary /usr/local/lib/python3.10/dist-packages/bitsandbytes/libbitsandbytes_cuda117.so...\n"
|
61 |
+
]
|
62 |
+
},
|
63 |
+
{
|
64 |
+
"name": "stderr",
|
65 |
+
"output_type": "stream",
|
66 |
+
"text": [
|
67 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('/usr/local/nvidia/lib64'), PosixPath('/usr/local/nvidia/lib')}\n",
|
68 |
+
" warn(msg)\n",
|
69 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: /usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/lib/x86_64-linux-gnu did not contain libcudart.so as expected! Searching further paths...\n",
|
70 |
+
" warn(msg)\n",
|
71 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('yCIZccJum+9/ZLm7c7rDHLQYnn/OPUl4MSQ9SgINSks'), PosixPath('SHA256')}\n",
|
72 |
+
" warn(msg)\n",
|
73 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('noninteractive SHELL=/bin/bash')}\n",
|
74 |
+
" warn(msg)\n",
|
75 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('/workspace/Untitled.ipynb')}\n",
|
76 |
+
" warn(msg)\n",
|
77 |
+
"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('//matplotlib_inline.backend_inline'), PosixPath('module')}\n",
|
78 |
+
" warn(msg)\n"
|
79 |
+
]
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"data": {
|
83 |
+
"application/vnd.jupyter.widget-view+json": {
|
84 |
+
"model_id": "5861e63248924916a4e574e3b559d5f9",
|
85 |
+
"version_major": 2,
|
86 |
+
"version_minor": 0
|
87 |
+
},
|
88 |
+
"text/plain": [
|
89 |
+
"Loading checkpoint shards: 0%| | 0/2 [00:00<?, ?it/s]"
|
90 |
+
]
|
91 |
+
},
|
92 |
+
"metadata": {},
|
93 |
+
"output_type": "display_data"
|
94 |
+
}
|
95 |
+
],
|
96 |
+
"source": [
|
97 |
+
"import torch\n",
|
98 |
+
"from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig\n",
|
99 |
+
"\n",
|
100 |
+
"model_id = \"/workspace/WizardLM-7b\"\n",
|
101 |
+
"bnb_config = BitsAndBytesConfig(\n",
|
102 |
+
" load_in_8bit=True,\n",
|
103 |
+
" bnb_8bit_use_double_quant=True,\n",
|
104 |
+
" bnb_8bit_quant_type=\"nf4\",\n",
|
105 |
+
" bnb_8bit_compute_dtype=torch.float32,\n",
|
106 |
+
" llm_int8_enable_fp32_cpu_offload=True,\n",
|
107 |
+
" load_in_8bit_fp32_cpu_offload=True\n",
|
108 |
+
")\n",
|
109 |
+
"\n",
|
110 |
+
"\n",
|
111 |
+
"\n",
|
112 |
+
"\n",
|
113 |
+
"tokenizer = AutoTokenizer.from_pretrained(model_id)\n",
|
114 |
+
"model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=bnb_config, device_map=\"auto\", max_memory={0:'8GiB', 'cpu':'50GiB'})"
|
115 |
+
]
|
116 |
+
},
|
117 |
+
{
|
118 |
+
"cell_type": "code",
|
119 |
+
"execution_count": 4,
|
120 |
+
"id": "6e575c22-66de-457c-82b2-da57db38ead8",
|
121 |
+
"metadata": {},
|
122 |
+
"outputs": [],
|
123 |
+
"source": [
|
124 |
+
"from peft import prepare_model_for_int8_training\n",
|
125 |
+
"\n",
|
126 |
+
"model.gradient_checkpointing_enable()\n",
|
127 |
+
"model = prepare_model_for_int8_training(model)"
|
128 |
+
]
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"cell_type": "code",
|
132 |
+
"execution_count": 5,
|
133 |
+
"id": "16fd3d91-6468-4268-bc48-3b1c5bcf39cc",
|
134 |
+
"metadata": {},
|
135 |
+
"outputs": [],
|
136 |
+
"source": [
|
137 |
+
"def print_trainable_parameters(model):\n",
|
138 |
+
" \"\"\"\n",
|
139 |
+
" Prints the number of trainable parameters in the model.\n",
|
140 |
+
" \"\"\"\n",
|
141 |
+
" trainable_params = 0\n",
|
142 |
+
" all_param = 0\n",
|
143 |
+
" for _, param in model.named_parameters():\n",
|
144 |
+
" all_param += param.numel()\n",
|
145 |
+
" if param.requires_grad:\n",
|
146 |
+
" trainable_params += param.numel()\n",
|
147 |
+
" print(\n",
|
148 |
+
" f\"trainable params: {trainable_params} || all params: {all_param} || trainable%: {100 * trainable_params / all_param}\"\n",
|
149 |
+
" )"
|
150 |
+
]
|
151 |
+
},
|
152 |
+
{
|
153 |
+
"cell_type": "code",
|
154 |
+
"execution_count": 6,
|
155 |
+
"id": "79566f41-ee3f-4662-bbcc-0c8ebc684b8d",
|
156 |
+
"metadata": {},
|
157 |
+
"outputs": [
|
158 |
+
{
|
159 |
+
"name": "stdout",
|
160 |
+
"output_type": "stream",
|
161 |
+
"text": [
|
162 |
+
"trainable params: 6291456 || all params: 6744707072 || trainable%: 0.09327989982127426\n"
|
163 |
+
]
|
164 |
+
}
|
165 |
+
],
|
166 |
+
"source": [
|
167 |
+
"from peft import LoraConfig, get_peft_model\n",
|
168 |
+
"\n",
|
169 |
+
"\n",
|
170 |
+
"config = LoraConfig(\n",
|
171 |
+
" r=8,\n",
|
172 |
+
" lora_alpha=64,\n",
|
173 |
+
" target_modules = [\"q_proj\", \"k_proj\", \"v_proj\", \"out_proj\", \"fc_in\", \"fc_out\", \"wte\"],\n",
|
174 |
+
" lora_dropout=0.1,\n",
|
175 |
+
" bias=\"all\",\n",
|
176 |
+
" task_type=\"CAUSAL_LM\"\n",
|
177 |
+
")\n",
|
178 |
+
"\n",
|
179 |
+
"model = get_peft_model(model, config)\n",
|
180 |
+
"print_trainable_parameters(model)\n"
|
181 |
+
]
|
182 |
+
},
|
183 |
+
{
|
184 |
+
"cell_type": "code",
|
185 |
+
"execution_count": 7,
|
186 |
+
"id": "68ca2e1a-eeca-41a0-a220-f27971309477",
|
187 |
+
"metadata": {},
|
188 |
+
"outputs": [
|
189 |
+
{
|
190 |
+
"name": "stderr",
|
191 |
+
"output_type": "stream",
|
192 |
+
"text": [
|
193 |
+
"Found cached dataset json (/root/.cache/huggingface/datasets/json/default-6d1f9620cbe5783f/0.0.0/8bb11242116d547c741b2e8a1f18598ffdd40a1d4f2a2872c7a28b697434bc96)\n",
|
194 |
+
"Loading cached processed dataset at /root/.cache/huggingface/datasets/json/default-6d1f9620cbe5783f/0.0.0/8bb11242116d547c741b2e8a1f18598ffdd40a1d4f2a2872c7a28b697434bc96/cache-6b78e4d017496646.arrow\n"
|
195 |
+
]
|
196 |
+
}
|
197 |
+
],
|
198 |
+
"source": [
|
199 |
+
"from datasets import load_dataset\n",
|
200 |
+
"\n",
|
201 |
+
"# Define a function to tokenize the dialogues\n",
|
202 |
+
"def tokenize_function(examples):\n",
|
203 |
+
" return tokenizer(examples[\"dialogue\"], truncation=True, padding='max_length', max_length=128)\n",
|
204 |
+
"\n",
|
205 |
+
"# Load the new dataset\n",
|
206 |
+
"dataset = load_dataset('json', data_files='/workspace/new_output.jsonl', split='train')\n",
|
207 |
+
"\n",
|
208 |
+
"# Tokenize the dataset\n",
|
209 |
+
"dataset = dataset.map(tokenize_function, batched=True)\n",
|
210 |
+
"\n",
|
211 |
+
"# Split the dataset into training and evaluation sets\n",
|
212 |
+
"dataset = dataset.train_test_split(test_size=0.1)\n",
|
213 |
+
"\n",
|
214 |
+
"# Get the training and evaluation datasets\n",
|
215 |
+
"train_dataset = dataset['train']\n",
|
216 |
+
"eval_dataset = dataset['test']"
|
217 |
+
]
|
218 |
+
},
|
219 |
+
{
|
220 |
+
"cell_type": "code",
|
221 |
+
"execution_count": null,
|
222 |
+
"id": "0ba14b1b-26ec-420c-bb41-6dacfaa5093a",
|
223 |
+
"metadata": {},
|
224 |
+
"outputs": [
|
225 |
+
{
|
226 |
+
"name": "stderr",
|
227 |
+
"output_type": "stream",
|
228 |
+
"text": [
|
229 |
+
"\u001b[34m\u001b[1mwandb\u001b[0m: Currently logged in as: \u001b[33mdj1150277539\u001b[0m. Use \u001b[1m`wandb login --relogin`\u001b[0m to force relogin\n"
|
230 |
+
]
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"data": {
|
234 |
+
"text/html": [
|
235 |
+
"Tracking run with wandb version 0.15.4"
|
236 |
+
],
|
237 |
+
"text/plain": [
|
238 |
+
"<IPython.core.display.HTML object>"
|
239 |
+
]
|
240 |
+
},
|
241 |
+
"metadata": {},
|
242 |
+
"output_type": "display_data"
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"data": {
|
246 |
+
"text/html": [
|
247 |
+
"Run data is saved locally in <code>/workspace/wandb/run-20230704_050553-7zgg0kn5</code>"
|
248 |
+
],
|
249 |
+
"text/plain": [
|
250 |
+
"<IPython.core.display.HTML object>"
|
251 |
+
]
|
252 |
+
},
|
253 |
+
"metadata": {},
|
254 |
+
"output_type": "display_data"
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"data": {
|
258 |
+
"text/html": [
|
259 |
+
"Syncing run <strong><a href='https://wandb.ai/dj1150277539/huggingface/runs/7zgg0kn5' target=\"_blank\">distinctive-feather-33</a></strong> to <a href='https://wandb.ai/dj1150277539/huggingface' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/run' target=\"_blank\">docs</a>)<br/>"
|
260 |
+
],
|
261 |
+
"text/plain": [
|
262 |
+
"<IPython.core.display.HTML object>"
|
263 |
+
]
|
264 |
+
},
|
265 |
+
"metadata": {},
|
266 |
+
"output_type": "display_data"
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"data": {
|
270 |
+
"text/html": [
|
271 |
+
" View project at <a href='https://wandb.ai/dj1150277539/huggingface' target=\"_blank\">https://wandb.ai/dj1150277539/huggingface</a>"
|
272 |
+
],
|
273 |
+
"text/plain": [
|
274 |
+
"<IPython.core.display.HTML object>"
|
275 |
+
]
|
276 |
+
},
|
277 |
+
"metadata": {},
|
278 |
+
"output_type": "display_data"
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"data": {
|
282 |
+
"text/html": [
|
283 |
+
" View run at <a href='https://wandb.ai/dj1150277539/huggingface/runs/7zgg0kn5' target=\"_blank\">https://wandb.ai/dj1150277539/huggingface/runs/7zgg0kn5</a>"
|
284 |
+
],
|
285 |
+
"text/plain": [
|
286 |
+
"<IPython.core.display.HTML object>"
|
287 |
+
]
|
288 |
+
},
|
289 |
+
"metadata": {},
|
290 |
+
"output_type": "display_data"
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"name": "stderr",
|
294 |
+
"output_type": "stream",
|
295 |
+
"text": [
|
296 |
+
"You're using a LlamaTokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding.\n"
|
297 |
+
]
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"data": {
|
301 |
+
"text/html": [
|
302 |
+
"\n",
|
303 |
+
" <div>\n",
|
304 |
+
" \n",
|
305 |
+
" <progress value='292' max='300' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
|
306 |
+
" [292/300 13:37:21 < 22:32, 0.01 it/s, Epoch 166/300]\n",
|
307 |
+
" </div>\n",
|
308 |
+
" <table border=\"1\" class=\"dataframe\">\n",
|
309 |
+
" <thead>\n",
|
310 |
+
" <tr style=\"text-align: left;\">\n",
|
311 |
+
" <th>Step</th>\n",
|
312 |
+
" <th>Training Loss</th>\n",
|
313 |
+
" <th>Validation Loss</th>\n",
|
314 |
+
" </tr>\n",
|
315 |
+
" </thead>\n",
|
316 |
+
" <tbody>\n",
|
317 |
+
" <tr>\n",
|
318 |
+
" <td>2</td>\n",
|
319 |
+
" <td>3.336500</td>\n",
|
320 |
+
" <td>14.149993</td>\n",
|
321 |
+
" </tr>\n",
|
322 |
+
" <tr>\n",
|
323 |
+
" <td>4</td>\n",
|
324 |
+
" <td>6.517100</td>\n",
|
325 |
+
" <td>13.468672</td>\n",
|
326 |
+
" </tr>\n",
|
327 |
+
" <tr>\n",
|
328 |
+
" <td>6</td>\n",
|
329 |
+
" <td>9.206500</td>\n",
|
330 |
+
" <td>12.650669</td>\n",
|
331 |
+
" </tr>\n",
|
332 |
+
" <tr>\n",
|
333 |
+
" <td>8</td>\n",
|
334 |
+
" <td>9.328000</td>\n",
|
335 |
+
" <td>11.601873</td>\n",
|
336 |
+
" </tr>\n",
|
337 |
+
" <tr>\n",
|
338 |
+
" <td>10</td>\n",
|
339 |
+
" <td>6.599700</td>\n",
|
340 |
+
" <td>10.404304</td>\n",
|
341 |
+
" </tr>\n",
|
342 |
+
" <tr>\n",
|
343 |
+
" <td>12</td>\n",
|
344 |
+
" <td>3.456600</td>\n",
|
345 |
+
" <td>8.535113</td>\n",
|
346 |
+
" </tr>\n",
|
347 |
+
" <tr>\n",
|
348 |
+
" <td>14</td>\n",
|
349 |
+
" <td>0.919500</td>\n",
|
350 |
+
" <td>6.562298</td>\n",
|
351 |
+
" </tr>\n",
|
352 |
+
" <tr>\n",
|
353 |
+
" <td>16</td>\n",
|
354 |
+
" <td>0.691900</td>\n",
|
355 |
+
" <td>6.245927</td>\n",
|
356 |
+
" </tr>\n",
|
357 |
+
" <tr>\n",
|
358 |
+
" <td>18</td>\n",
|
359 |
+
" <td>1.991800</td>\n",
|
360 |
+
" <td>5.460762</td>\n",
|
361 |
+
" </tr>\n",
|
362 |
+
" <tr>\n",
|
363 |
+
" <td>20</td>\n",
|
364 |
+
" <td>2.966800</td>\n",
|
365 |
+
" <td>4.838455</td>\n",
|
366 |
+
" </tr>\n",
|
367 |
+
" <tr>\n",
|
368 |
+
" <td>22</td>\n",
|
369 |
+
" <td>3.540500</td>\n",
|
370 |
+
" <td>4.411865</td>\n",
|
371 |
+
" </tr>\n",
|
372 |
+
" <tr>\n",
|
373 |
+
" <td>24</td>\n",
|
374 |
+
" <td>3.101700</td>\n",
|
375 |
+
" <td>4.204891</td>\n",
|
376 |
+
" </tr>\n",
|
377 |
+
" <tr>\n",
|
378 |
+
" <td>26</td>\n",
|
379 |
+
" <td>1.990200</td>\n",
|
380 |
+
" <td>4.001011</td>\n",
|
381 |
+
" </tr>\n",
|
382 |
+
" <tr>\n",
|
383 |
+
" <td>28</td>\n",
|
384 |
+
" <td>0.980100</td>\n",
|
385 |
+
" <td>3.855425</td>\n",
|
386 |
+
" </tr>\n",
|
387 |
+
" <tr>\n",
|
388 |
+
" <td>30</td>\n",
|
389 |
+
" <td>0.058200</td>\n",
|
390 |
+
" <td>3.714956</td>\n",
|
391 |
+
" </tr>\n",
|
392 |
+
" <tr>\n",
|
393 |
+
" <td>32</td>\n",
|
394 |
+
" <td>0.800900</td>\n",
|
395 |
+
" <td>3.611897</td>\n",
|
396 |
+
" </tr>\n",
|
397 |
+
" <tr>\n",
|
398 |
+
" <td>34</td>\n",
|
399 |
+
" <td>1.613400</td>\n",
|
400 |
+
" <td>3.506055</td>\n",
|
401 |
+
" </tr>\n",
|
402 |
+
" <tr>\n",
|
403 |
+
" <td>36</td>\n",
|
404 |
+
" <td>2.387300</td>\n",
|
405 |
+
" <td>3.416133</td>\n",
|
406 |
+
" </tr>\n",
|
407 |
+
" <tr>\n",
|
408 |
+
" <td>38</td>\n",
|
409 |
+
" <td>2.598900</td>\n",
|
410 |
+
" <td>3.339494</td>\n",
|
411 |
+
" </tr>\n",
|
412 |
+
" <tr>\n",
|
413 |
+
" <td>40</td>\n",
|
414 |
+
" <td>2.018800</td>\n",
|
415 |
+
" <td>3.274912</td>\n",
|
416 |
+
" </tr>\n",
|
417 |
+
" <tr>\n",
|
418 |
+
" <td>42</td>\n",
|
419 |
+
" <td>1.217900</td>\n",
|
420 |
+
" <td>3.217013</td>\n",
|
421 |
+
" </tr>\n",
|
422 |
+
" <tr>\n",
|
423 |
+
" <td>44</td>\n",
|
424 |
+
" <td>0.448600</td>\n",
|
425 |
+
" <td>3.172593</td>\n",
|
426 |
+
" </tr>\n",
|
427 |
+
" <tr>\n",
|
428 |
+
" <td>46</td>\n",
|
429 |
+
" <td>0.293700</td>\n",
|
430 |
+
" <td>3.141093</td>\n",
|
431 |
+
" </tr>\n",
|
432 |
+
" <tr>\n",
|
433 |
+
" <td>48</td>\n",
|
434 |
+
" <td>1.022200</td>\n",
|
435 |
+
" <td>3.091443</td>\n",
|
436 |
+
" </tr>\n",
|
437 |
+
" <tr>\n",
|
438 |
+
" <td>50</td>\n",
|
439 |
+
" <td>1.731100</td>\n",
|
440 |
+
" <td>3.060865</td>\n",
|
441 |
+
" </tr>\n",
|
442 |
+
" <tr>\n",
|
443 |
+
" <td>52</td>\n",
|
444 |
+
" <td>2.338500</td>\n",
|
445 |
+
" <td>3.032907</td>\n",
|
446 |
+
" </tr>\n",
|
447 |
+
" <tr>\n",
|
448 |
+
" <td>54</td>\n",
|
449 |
+
" <td>2.226600</td>\n",
|
450 |
+
" <td>3.012452</td>\n",
|
451 |
+
" </tr>\n",
|
452 |
+
" <tr>\n",
|
453 |
+
" <td>56</td>\n",
|
454 |
+
" <td>1.500500</td>\n",
|
455 |
+
" <td>2.992773</td>\n",
|
456 |
+
" </tr>\n",
|
457 |
+
" <tr>\n",
|
458 |
+
" <td>58</td>\n",
|
459 |
+
" <td>0.795400</td>\n",
|
460 |
+
" <td>2.976812</td>\n",
|
461 |
+
" </tr>\n",
|
462 |
+
" <tr>\n",
|
463 |
+
" <td>60</td>\n",
|
464 |
+
" <td>0.092900</td>\n",
|
465 |
+
" <td>2.965430</td>\n",
|
466 |
+
" </tr>\n",
|
467 |
+
" <tr>\n",
|
468 |
+
" <td>62</td>\n",
|
469 |
+
" <td>0.597500</td>\n",
|
470 |
+
" <td>2.950096</td>\n",
|
471 |
+
" </tr>\n",
|
472 |
+
" <tr>\n",
|
473 |
+
" <td>64</td>\n",
|
474 |
+
" <td>1.287900</td>\n",
|
475 |
+
" <td>2.940890</td>\n",
|
476 |
+
" </tr>\n",
|
477 |
+
" <tr>\n",
|
478 |
+
" <td>66</td>\n",
|
479 |
+
" <td>1.970100</td>\n",
|
480 |
+
" <td>2.930588</td>\n",
|
481 |
+
" </tr>\n",
|
482 |
+
" <tr>\n",
|
483 |
+
" <td>68</td>\n",
|
484 |
+
" <td>2.240600</td>\n",
|
485 |
+
" <td>2.923621</td>\n",
|
486 |
+
" </tr>\n",
|
487 |
+
" <tr>\n",
|
488 |
+
" <td>70</td>\n",
|
489 |
+
" <td>1.825200</td>\n",
|
490 |
+
" <td>2.917476</td>\n",
|
491 |
+
" </tr>\n",
|
492 |
+
" <tr>\n",
|
493 |
+
" <td>72</td>\n",
|
494 |
+
" <td>1.137100</td>\n",
|
495 |
+
" <td>2.911416</td>\n",
|
496 |
+
" </tr>\n",
|
497 |
+
" <tr>\n",
|
498 |
+
" <td>74</td>\n",
|
499 |
+
" <td>0.455000</td>\n",
|
500 |
+
" <td>2.906709</td>\n",
|
501 |
+
" </tr>\n",
|
502 |
+
" <tr>\n",
|
503 |
+
" <td>76</td>\n",
|
504 |
+
" <td>0.225000</td>\n",
|
505 |
+
" <td>2.902960</td>\n",
|
506 |
+
" </tr>\n",
|
507 |
+
" <tr>\n",
|
508 |
+
" <td>78</td>\n",
|
509 |
+
" <td>0.907000</td>\n",
|
510 |
+
" <td>2.898351</td>\n",
|
511 |
+
" </tr>\n",
|
512 |
+
" <tr>\n",
|
513 |
+
" <td>80</td>\n",
|
514 |
+
" <td>1.582700</td>\n",
|
515 |
+
" <td>2.894657</td>\n",
|
516 |
+
" </tr>\n",
|
517 |
+
" <tr>\n",
|
518 |
+
" <td>82</td>\n",
|
519 |
+
" <td>2.208000</td>\n",
|
520 |
+
" <td>2.891902</td>\n",
|
521 |
+
" </tr>\n",
|
522 |
+
" <tr>\n",
|
523 |
+
" <td>84</td>\n",
|
524 |
+
" <td>2.161300</td>\n",
|
525 |
+
" <td>2.889836</td>\n",
|
526 |
+
" </tr>\n",
|
527 |
+
" <tr>\n",
|
528 |
+
" <td>86</td>\n",
|
529 |
+
" <td>1.485300</td>\n",
|
530 |
+
" <td>2.886881</td>\n",
|
531 |
+
" </tr>\n",
|
532 |
+
" <tr>\n",
|
533 |
+
" <td>88</td>\n",
|
534 |
+
" <td>0.808700</td>\n",
|
535 |
+
" <td>2.888359</td>\n",
|
536 |
+
" </tr>\n",
|
537 |
+
" <tr>\n",
|
538 |
+
" <td>90</td>\n",
|
539 |
+
" <td>0.134600</td>\n",
|
540 |
+
" <td>2.898380</td>\n",
|
541 |
+
" </tr>\n",
|
542 |
+
" <tr>\n",
|
543 |
+
" <td>92</td>\n",
|
544 |
+
" <td>0.540100</td>\n",
|
545 |
+
" <td>2.904707</td>\n",
|
546 |
+
" </tr>\n",
|
547 |
+
" <tr>\n",
|
548 |
+
" <td>94</td>\n",
|
549 |
+
" <td>1.212500</td>\n",
|
550 |
+
" <td>2.884937</td>\n",
|
551 |
+
" </tr>\n",
|
552 |
+
" <tr>\n",
|
553 |
+
" <td>96</td>\n",
|
554 |
+
" <td>1.882300</td>\n",
|
555 |
+
" <td>2.886901</td>\n",
|
556 |
+
" </tr>\n",
|
557 |
+
" <tr>\n",
|
558 |
+
" <td>98</td>\n",
|
559 |
+
" <td>2.190800</td>\n",
|
560 |
+
" <td>2.881487</td>\n",
|
561 |
+
" </tr>\n",
|
562 |
+
" <tr>\n",
|
563 |
+
" <td>100</td>\n",
|
564 |
+
" <td>1.831900</td>\n",
|
565 |
+
" <td>2.881773</td>\n",
|
566 |
+
" </tr>\n",
|
567 |
+
" <tr>\n",
|
568 |
+
" <td>102</td>\n",
|
569 |
+
" <td>1.164300</td>\n",
|
570 |
+
" <td>2.879683</td>\n",
|
571 |
+
" </tr>\n",
|
572 |
+
" <tr>\n",
|
573 |
+
" <td>104</td>\n",
|
574 |
+
" <td>0.490800</td>\n",
|
575 |
+
" <td>2.878468</td>\n",
|
576 |
+
" </tr>\n",
|
577 |
+
" <tr>\n",
|
578 |
+
" <td>106</td>\n",
|
579 |
+
" <td>0.178100</td>\n",
|
580 |
+
" <td>2.877084</td>\n",
|
581 |
+
" </tr>\n",
|
582 |
+
" <tr>\n",
|
583 |
+
" <td>108</td>\n",
|
584 |
+
" <td>0.841800</td>\n",
|
585 |
+
" <td>2.875591</td>\n",
|
586 |
+
" </tr>\n",
|
587 |
+
" <tr>\n",
|
588 |
+
" <td>110</td>\n",
|
589 |
+
" <td>1.500900</td>\n",
|
590 |
+
" <td>2.874607</td>\n",
|
591 |
+
" </tr>\n",
|
592 |
+
" <tr>\n",
|
593 |
+
" <td>112</td>\n",
|
594 |
+
" <td>2.174100</td>\n",
|
595 |
+
" <td>2.873750</td>\n",
|
596 |
+
" </tr>\n",
|
597 |
+
" <tr>\n",
|
598 |
+
" <td>114</td>\n",
|
599 |
+
" <td>0.664800</td>\n",
|
600 |
+
" <td>2.875177</td>\n",
|
601 |
+
" </tr>\n",
|
602 |
+
" <tr>\n",
|
603 |
+
" <td>116</td>\n",
|
604 |
+
" <td>1.330800</td>\n",
|
605 |
+
" <td>2.873417</td>\n",
|
606 |
+
" </tr>\n",
|
607 |
+
" <tr>\n",
|
608 |
+
" <td>118</td>\n",
|
609 |
+
" <td>1.988000</td>\n",
|
610 |
+
" <td>2.872393</td>\n",
|
611 |
+
" </tr>\n",
|
612 |
+
" <tr>\n",
|
613 |
+
" <td>120</td>\n",
|
614 |
+
" <td>2.162500</td>\n",
|
615 |
+
" <td>2.873486</td>\n",
|
616 |
+
" </tr>\n",
|
617 |
+
" <tr>\n",
|
618 |
+
" <td>122</td>\n",
|
619 |
+
" <td>1.675900</td>\n",
|
620 |
+
" <td>2.873488</td>\n",
|
621 |
+
" </tr>\n",
|
622 |
+
" <tr>\n",
|
623 |
+
" <td>124</td>\n",
|
624 |
+
" <td>1.014600</td>\n",
|
625 |
+
" <td>2.874382</td>\n",
|
626 |
+
" </tr>\n",
|
627 |
+
" <tr>\n",
|
628 |
+
" <td>126</td>\n",
|
629 |
+
" <td>0.350600</td>\n",
|
630 |
+
" <td>2.878257</td>\n",
|
631 |
+
" </tr>\n",
|
632 |
+
" <tr>\n",
|
633 |
+
" <td>128</td>\n",
|
634 |
+
" <td>0.307900</td>\n",
|
635 |
+
" <td>2.882156</td>\n",
|
636 |
+
" </tr>\n",
|
637 |
+
" <tr>\n",
|
638 |
+
" <td>130</td>\n",
|
639 |
+
" <td>0.963200</td>\n",
|
640 |
+
" <td>2.874237</td>\n",
|
641 |
+
" </tr>\n",
|
642 |
+
" <tr>\n",
|
643 |
+
" <td>132</td>\n",
|
644 |
+
" <td>1.624300</td>\n",
|
645 |
+
" <td>2.872886</td>\n",
|
646 |
+
" </tr>\n",
|
647 |
+
" <tr>\n",
|
648 |
+
" <td>134</td>\n",
|
649 |
+
" <td>2.146800</td>\n",
|
650 |
+
" <td>2.873395</td>\n",
|
651 |
+
" </tr>\n",
|
652 |
+
" <tr>\n",
|
653 |
+
" <td>136</td>\n",
|
654 |
+
" <td>2.014700</td>\n",
|
655 |
+
" <td>2.875467</td>\n",
|
656 |
+
" </tr>\n",
|
657 |
+
" <tr>\n",
|
658 |
+
" <td>138</td>\n",
|
659 |
+
" <td>1.351300</td>\n",
|
660 |
+
" <td>2.876896</td>\n",
|
661 |
+
" </tr>\n",
|
662 |
+
" <tr>\n",
|
663 |
+
" <td>140</td>\n",
|
664 |
+
" <td>0.699800</td>\n",
|
665 |
+
" <td>2.879110</td>\n",
|
666 |
+
" </tr>\n",
|
667 |
+
" <tr>\n",
|
668 |
+
" <td>142</td>\n",
|
669 |
+
" <td>0.044100</td>\n",
|
670 |
+
" <td>2.881222</td>\n",
|
671 |
+
" </tr>\n",
|
672 |
+
" <tr>\n",
|
673 |
+
" <td>144</td>\n",
|
674 |
+
" <td>0.610900</td>\n",
|
675 |
+
" <td>2.880388</td>\n",
|
676 |
+
" </tr>\n",
|
677 |
+
" <tr>\n",
|
678 |
+
" <td>146</td>\n",
|
679 |
+
" <td>1.260600</td>\n",
|
680 |
+
" <td>2.882942</td>\n",
|
681 |
+
" </tr>\n",
|
682 |
+
" <tr>\n",
|
683 |
+
" <td>148</td>\n",
|
684 |
+
" <td>1.908200</td>\n",
|
685 |
+
" <td>2.885541</td>\n",
|
686 |
+
" </tr>\n",
|
687 |
+
" <tr>\n",
|
688 |
+
" <td>150</td>\n",
|
689 |
+
" <td>2.125100</td>\n",
|
690 |
+
" <td>2.886549</td>\n",
|
691 |
+
" </tr>\n",
|
692 |
+
" <tr>\n",
|
693 |
+
" <td>152</td>\n",
|
694 |
+
" <td>1.685200</td>\n",
|
695 |
+
" <td>2.889215</td>\n",
|
696 |
+
" </tr>\n",
|
697 |
+
" <tr>\n",
|
698 |
+
" <td>154</td>\n",
|
699 |
+
" <td>1.042300</td>\n",
|
700 |
+
" <td>2.893012</td>\n",
|
701 |
+
" </tr>\n",
|
702 |
+
" <tr>\n",
|
703 |
+
" <td>156</td>\n",
|
704 |
+
" <td>0.386300</td>\n",
|
705 |
+
" <td>2.897541</td>\n",
|
706 |
+
" </tr>\n",
|
707 |
+
" <tr>\n",
|
708 |
+
" <td>158</td>\n",
|
709 |
+
" <td>0.258700</td>\n",
|
710 |
+
" <td>2.897153</td>\n",
|
711 |
+
" </tr>\n",
|
712 |
+
" <tr>\n",
|
713 |
+
" <td>160</td>\n",
|
714 |
+
" <td>0.901400</td>\n",
|
715 |
+
" <td>2.905147</td>\n",
|
716 |
+
" </tr>\n",
|
717 |
+
" <tr>\n",
|
718 |
+
" <td>162</td>\n",
|
719 |
+
" <td>1.545900</td>\n",
|
720 |
+
" <td>2.902808</td>\n",
|
721 |
+
" </tr>\n",
|
722 |
+
" <tr>\n",
|
723 |
+
" <td>164</td>\n",
|
724 |
+
" <td>2.106000</td>\n",
|
725 |
+
" <td>2.907741</td>\n",
|
726 |
+
" </tr>\n",
|
727 |
+
" <tr>\n",
|
728 |
+
" <td>166</td>\n",
|
729 |
+
" <td>2.025800</td>\n",
|
730 |
+
" <td>2.897594</td>\n",
|
731 |
+
" </tr>\n",
|
732 |
+
" <tr>\n",
|
733 |
+
" <td>168</td>\n",
|
734 |
+
" <td>1.384000</td>\n",
|
735 |
+
" <td>2.900589</td>\n",
|
736 |
+
" </tr>\n",
|
737 |
+
" <tr>\n",
|
738 |
+
" <td>170</td>\n",
|
739 |
+
" <td>0.730600</td>\n",
|
740 |
+
" <td>2.906770</td>\n",
|
741 |
+
" </tr>\n",
|
742 |
+
" <tr>\n",
|
743 |
+
" <td>172</td>\n",
|
744 |
+
" <td>0.087300</td>\n",
|
745 |
+
" <td>2.923102</td>\n",
|
746 |
+
" </tr>\n",
|
747 |
+
" <tr>\n",
|
748 |
+
" <td>174</td>\n",
|
749 |
+
" <td>0.556200</td>\n",
|
750 |
+
" <td>2.896429</td>\n",
|
751 |
+
" </tr>\n",
|
752 |
+
" <tr>\n",
|
753 |
+
" <td>176</td>\n",
|
754 |
+
" <td>1.199200</td>\n",
|
755 |
+
" <td>2.916603</td>\n",
|
756 |
+
" </tr>\n",
|
757 |
+
" <tr>\n",
|
758 |
+
" <td>178</td>\n",
|
759 |
+
" <td>1.837500</td>\n",
|
760 |
+
" <td>2.915208</td>\n",
|
761 |
+
" </tr>\n",
|
762 |
+
" <tr>\n",
|
763 |
+
" <td>180</td>\n",
|
764 |
+
" <td>2.091800</td>\n",
|
765 |
+
" <td>2.918897</td>\n",
|
766 |
+
" </tr>\n",
|
767 |
+
" <tr>\n",
|
768 |
+
" <td>182</td>\n",
|
769 |
+
" <td>1.706300</td>\n",
|
770 |
+
" <td>2.923554</td>\n",
|
771 |
+
" </tr>\n",
|
772 |
+
" <tr>\n",
|
773 |
+
" <td>184</td>\n",
|
774 |
+
" <td>1.060900</td>\n",
|
775 |
+
" <td>2.924343</td>\n",
|
776 |
+
" </tr>\n",
|
777 |
+
" <tr>\n",
|
778 |
+
" <td>186</td>\n",
|
779 |
+
" <td>0.426200</td>\n",
|
780 |
+
" <td>2.937875</td>\n",
|
781 |
+
" </tr>\n",
|
782 |
+
" <tr>\n",
|
783 |
+
" <td>188</td>\n",
|
784 |
+
" <td>0.213300</td>\n",
|
785 |
+
" <td>2.928014</td>\n",
|
786 |
+
" </tr>\n",
|
787 |
+
" <tr>\n",
|
788 |
+
" <td>190</td>\n",
|
789 |
+
" <td>0.846600</td>\n",
|
790 |
+
" <td>2.946512</td>\n",
|
791 |
+
" </tr>\n",
|
792 |
+
" <tr>\n",
|
793 |
+
" <td>192</td>\n",
|
794 |
+
" <td>1.483100</td>\n",
|
795 |
+
" <td>2.939318</td>\n",
|
796 |
+
" </tr>\n",
|
797 |
+
" <tr>\n",
|
798 |
+
" <td>194</td>\n",
|
799 |
+
" <td>2.072100</td>\n",
|
800 |
+
" <td>2.945714</td>\n",
|
801 |
+
" </tr>\n",
|
802 |
+
" <tr>\n",
|
803 |
+
" <td>196</td>\n",
|
804 |
+
" <td>2.029300</td>\n",
|
805 |
+
" <td>2.949480</td>\n",
|
806 |
+
" </tr>\n",
|
807 |
+
" <tr>\n",
|
808 |
+
" <td>198</td>\n",
|
809 |
+
" <td>1.394200</td>\n",
|
810 |
+
" <td>2.952962</td>\n",
|
811 |
+
" </tr>\n",
|
812 |
+
" <tr>\n",
|
813 |
+
" <td>200</td>\n",
|
814 |
+
" <td>0.761500</td>\n",
|
815 |
+
" <td>2.967859</td>\n",
|
816 |
+
" </tr>\n",
|
817 |
+
" <tr>\n",
|
818 |
+
" <td>202</td>\n",
|
819 |
+
" <td>0.123500</td>\n",
|
820 |
+
" <td>2.950326</td>\n",
|
821 |
+
" </tr>\n",
|
822 |
+
" <tr>\n",
|
823 |
+
" <td>204</td>\n",
|
824 |
+
" <td>0.505700</td>\n",
|
825 |
+
" <td>2.971651</td>\n",
|
826 |
+
" </tr>\n",
|
827 |
+
" <tr>\n",
|
828 |
+
" <td>206</td>\n",
|
829 |
+
" <td>1.131400</td>\n",
|
830 |
+
" <td>2.973131</td>\n",
|
831 |
+
" </tr>\n",
|
832 |
+
" <tr>\n",
|
833 |
+
" <td>208</td>\n",
|
834 |
+
" <td>1.762000</td>\n",
|
835 |
+
" <td>2.972790</td>\n",
|
836 |
+
" </tr>\n",
|
837 |
+
" <tr>\n",
|
838 |
+
" <td>210</td>\n",
|
839 |
+
" <td>2.054900</td>\n",
|
840 |
+
" <td>2.977710</td>\n",
|
841 |
+
" </tr>\n",
|
842 |
+
" <tr>\n",
|
843 |
+
" <td>212</td>\n",
|
844 |
+
" <td>1.715500</td>\n",
|
845 |
+
" <td>2.984576</td>\n",
|
846 |
+
" </tr>\n",
|
847 |
+
" <tr>\n",
|
848 |
+
" <td>214</td>\n",
|
849 |
+
" <td>1.087700</td>\n",
|
850 |
+
" <td>2.982883</td>\n",
|
851 |
+
" </tr>\n",
|
852 |
+
" <tr>\n",
|
853 |
+
" <td>216</td>\n",
|
854 |
+
" <td>0.456900</td>\n",
|
855 |
+
" <td>2.988919</td>\n",
|
856 |
+
" </tr>\n",
|
857 |
+
" <tr>\n",
|
858 |
+
" <td>218</td>\n",
|
859 |
+
" <td>0.166900</td>\n",
|
860 |
+
" <td>2.996212</td>\n",
|
861 |
+
" </tr>\n",
|
862 |
+
" <tr>\n",
|
863 |
+
" <td>220</td>\n",
|
864 |
+
" <td>0.794400</td>\n",
|
865 |
+
" <td>2.994239</td>\n",
|
866 |
+
" </tr>\n",
|
867 |
+
" <tr>\n",
|
868 |
+
" <td>222</td>\n",
|
869 |
+
" <td>1.415700</td>\n",
|
870 |
+
" <td>3.001529</td>\n",
|
871 |
+
" </tr>\n",
|
872 |
+
" <tr>\n",
|
873 |
+
" <td>224</td>\n",
|
874 |
+
" <td>2.044000</td>\n",
|
875 |
+
" <td>3.003330</td>\n",
|
876 |
+
" </tr>\n",
|
877 |
+
" <tr>\n",
|
878 |
+
" <td>226</td>\n",
|
879 |
+
" <td>0.622300</td>\n",
|
880 |
+
" <td>3.006619</td>\n",
|
881 |
+
" </tr>\n",
|
882 |
+
" <tr>\n",
|
883 |
+
" <td>228</td>\n",
|
884 |
+
" <td>1.249300</td>\n",
|
885 |
+
" <td>3.011294</td>\n",
|
886 |
+
" </tr>\n",
|
887 |
+
" <tr>\n",
|
888 |
+
" <td>230</td>\n",
|
889 |
+
" <td>1.873000</td>\n",
|
890 |
+
" <td>3.013264</td>\n",
|
891 |
+
" </tr>\n",
|
892 |
+
" <tr>\n",
|
893 |
+
" <td>232</td>\n",
|
894 |
+
" <td>2.038300</td>\n",
|
895 |
+
" <td>3.014204</td>\n",
|
896 |
+
" </tr>\n",
|
897 |
+
" <tr>\n",
|
898 |
+
" <td>234</td>\n",
|
899 |
+
" <td>1.581300</td>\n",
|
900 |
+
" <td>3.021334</td>\n",
|
901 |
+
" </tr>\n",
|
902 |
+
" <tr>\n",
|
903 |
+
" <td>236</td>\n",
|
904 |
+
" <td>0.957400</td>\n",
|
905 |
+
" <td>3.019286</td>\n",
|
906 |
+
" </tr>\n",
|
907 |
+
" <tr>\n",
|
908 |
+
" <td>238</td>\n",
|
909 |
+
" <td>0.334200</td>\n",
|
910 |
+
" <td>3.023535</td>\n",
|
911 |
+
" </tr>\n",
|
912 |
+
" <tr>\n",
|
913 |
+
" <td>240</td>\n",
|
914 |
+
" <td>0.290700</td>\n",
|
915 |
+
" <td>3.026750</td>\n",
|
916 |
+
" </tr>\n",
|
917 |
+
" <tr>\n",
|
918 |
+
" <td>242</td>\n",
|
919 |
+
" <td>0.915000</td>\n",
|
920 |
+
" <td>3.029947</td>\n",
|
921 |
+
" </tr>\n",
|
922 |
+
" <tr>\n",
|
923 |
+
" <td>244</td>\n",
|
924 |
+
" <td>1.532900</td>\n",
|
925 |
+
" <td>3.030478</td>\n",
|
926 |
+
" </tr>\n",
|
927 |
+
" <tr>\n",
|
928 |
+
" <td>246</td>\n",
|
929 |
+
" <td>2.030900</td>\n",
|
930 |
+
" <td>3.033974</td>\n",
|
931 |
+
" </tr>\n",
|
932 |
+
" <tr>\n",
|
933 |
+
" <td>248</td>\n",
|
934 |
+
" <td>1.905700</td>\n",
|
935 |
+
" <td>3.042156</td>\n",
|
936 |
+
" </tr>\n",
|
937 |
+
" <tr>\n",
|
938 |
+
" <td>250</td>\n",
|
939 |
+
" <td>1.282600</td>\n",
|
940 |
+
" <td>3.039005</td>\n",
|
941 |
+
" </tr>\n",
|
942 |
+
" <tr>\n",
|
943 |
+
" <td>252</td>\n",
|
944 |
+
" <td>0.661500</td>\n",
|
945 |
+
" <td>3.036432</td>\n",
|
946 |
+
" </tr>\n",
|
947 |
+
" <tr>\n",
|
948 |
+
" <td>254</td>\n",
|
949 |
+
" <td>0.041400</td>\n",
|
950 |
+
" <td>3.039858</td>\n",
|
951 |
+
" </tr>\n",
|
952 |
+
" <tr>\n",
|
953 |
+
" <td>256</td>\n",
|
954 |
+
" <td>0.580900</td>\n",
|
955 |
+
" <td>3.042423</td>\n",
|
956 |
+
" </tr>\n",
|
957 |
+
" <tr>\n",
|
958 |
+
" <td>258</td>\n",
|
959 |
+
" <td>1.199400</td>\n",
|
960 |
+
" <td>3.042602</td>\n",
|
961 |
+
" </tr>\n",
|
962 |
+
" <tr>\n",
|
963 |
+
" <td>260</td>\n",
|
964 |
+
" <td>1.818700</td>\n",
|
965 |
+
" <td>3.046433</td>\n",
|
966 |
+
" </tr>\n",
|
967 |
+
" <tr>\n",
|
968 |
+
" <td>262</td>\n",
|
969 |
+
" <td>2.023600</td>\n",
|
970 |
+
" <td>3.047827</td>\n",
|
971 |
+
" </tr>\n",
|
972 |
+
" <tr>\n",
|
973 |
+
" <td>264</td>\n",
|
974 |
+
" <td>1.608300</td>\n",
|
975 |
+
" <td>3.050818</td>\n",
|
976 |
+
" </tr>\n",
|
977 |
+
" <tr>\n",
|
978 |
+
" <td>266</td>\n",
|
979 |
+
" <td>0.993300</td>\n",
|
980 |
+
" <td>3.050040</td>\n",
|
981 |
+
" </tr>\n",
|
982 |
+
" <tr>\n",
|
983 |
+
" <td>268</td>\n",
|
984 |
+
" <td>0.371800</td>\n",
|
985 |
+
" <td>3.053066</td>\n",
|
986 |
+
" </tr>\n",
|
987 |
+
" <tr>\n",
|
988 |
+
" <td>270</td>\n",
|
989 |
+
" <td>0.247700</td>\n",
|
990 |
+
" <td>3.051264</td>\n",
|
991 |
+
" </tr>\n",
|
992 |
+
" <tr>\n",
|
993 |
+
" <td>272</td>\n",
|
994 |
+
" <td>0.863300</td>\n",
|
995 |
+
" <td>3.050924</td>\n",
|
996 |
+
" </tr>\n",
|
997 |
+
" <tr>\n",
|
998 |
+
" <td>274</td>\n",
|
999 |
+
" <td>1.487000</td>\n",
|
1000 |
+
" <td>3.053752</td>\n",
|
1001 |
+
" </tr>\n",
|
1002 |
+
" <tr>\n",
|
1003 |
+
" <td>276</td>\n",
|
1004 |
+
" <td>2.020500</td>\n",
|
1005 |
+
" <td>3.056072</td>\n",
|
1006 |
+
" </tr>\n",
|
1007 |
+
" <tr>\n",
|
1008 |
+
" <td>278</td>\n",
|
1009 |
+
" <td>1.937400</td>\n",
|
1010 |
+
" <td>3.054851</td>\n",
|
1011 |
+
" </tr>\n",
|
1012 |
+
" <tr>\n",
|
1013 |
+
" <td>280</td>\n",
|
1014 |
+
" <td>1.321400</td>\n",
|
1015 |
+
" <td>3.053991</td>\n",
|
1016 |
+
" </tr>\n",
|
1017 |
+
" <tr>\n",
|
1018 |
+
" <td>282</td>\n",
|
1019 |
+
" <td>0.704000</td>\n",
|
1020 |
+
" <td>3.053047</td>\n",
|
1021 |
+
" </tr>\n",
|
1022 |
+
" <tr>\n",
|
1023 |
+
" <td>284</td>\n",
|
1024 |
+
" <td>0.082900</td>\n",
|
1025 |
+
" <td>3.054484</td>\n",
|
1026 |
+
" </tr>\n",
|
1027 |
+
" <tr>\n",
|
1028 |
+
" <td>286</td>\n",
|
1029 |
+
" <td>0.536800</td>\n",
|
1030 |
+
" <td>3.054063</td>\n",
|
1031 |
+
" </tr>\n",
|
1032 |
+
" <tr>\n",
|
1033 |
+
" <td>288</td>\n",
|
1034 |
+
" <td>1.155600</td>\n",
|
1035 |
+
" <td>3.050694</td>\n",
|
1036 |
+
" </tr>\n",
|
1037 |
+
" <tr>\n",
|
1038 |
+
" <td>290</td>\n",
|
1039 |
+
" <td>1.777500</td>\n",
|
1040 |
+
" <td>3.054609</td>\n",
|
1041 |
+
" </tr>\n",
|
1042 |
+
" </tbody>\n",
|
1043 |
+
"</table><p>"
|
1044 |
+
],
|
1045 |
+
"text/plain": [
|
1046 |
+
"<IPython.core.display.HTML object>"
|
1047 |
+
]
|
1048 |
+
},
|
1049 |
+
"metadata": {},
|
1050 |
+
"output_type": "display_data"
|
1051 |
+
}
|
1052 |
+
],
|
1053 |
+
"source": [
|
1054 |
+
"import transformers\n",
|
1055 |
+
"\n",
|
1056 |
+
"# needed for gpt-neo-x tokenizer\n",
|
1057 |
+
"tokenizer.pad_token = tokenizer.eos_token\n",
|
1058 |
+
"\n",
|
1059 |
+
"trainer = transformers.Trainer(\n",
|
1060 |
+
" model=model,\n",
|
1061 |
+
" train_dataset=train_dataset,\n",
|
1062 |
+
" eval_dataset=eval_dataset,\n",
|
1063 |
+
" args=transformers.TrainingArguments(\n",
|
1064 |
+
" evaluation_strategy=\"steps\",\n",
|
1065 |
+
" per_device_train_batch_size=64,\n",
|
1066 |
+
" per_device_eval_batch_size=16,\n",
|
1067 |
+
" gradient_accumulation_steps=64,\n",
|
1068 |
+
" warmup_steps=30,\n",
|
1069 |
+
" save_steps=100,\n",
|
1070 |
+
" eval_steps=2,\n",
|
1071 |
+
" num_train_epochs=300,\n",
|
1072 |
+
" learning_rate=3e-4,\n",
|
1073 |
+
" save_total_limit=2,\n",
|
1074 |
+
" load_best_model_at_end= True,\n",
|
1075 |
+
" fp16=True,\n",
|
1076 |
+
" logging_steps=1,\n",
|
1077 |
+
" lr_scheduler_type=\"cosine\",\n",
|
1078 |
+
" output_dir=\"/workspace\",\n",
|
1079 |
+
" optim=\"adamw_torch\",\n",
|
1080 |
+
" ),\n",
|
1081 |
+
" data_collator=transformers.DataCollatorForLanguageModeling(tokenizer, mlm=False),\n",
|
1082 |
+
")\n",
|
1083 |
+
"model.config.use_cache = False # silence the warnings. Please re-enable for inference!\n",
|
1084 |
+
"trainer.train()"
|
1085 |
+
]
|
1086 |
+
},
|
1087 |
+
{
|
1088 |
+
"cell_type": "code",
|
1089 |
+
"execution_count": 9,
|
1090 |
+
"id": "c6be457c-c54b-4191-885a-475745c75293",
|
1091 |
+
"metadata": {},
|
1092 |
+
"outputs": [],
|
1093 |
+
"source": [
|
1094 |
+
"#saving the adapter model\n",
|
1095 |
+
"pt_save_directory = '/workspace'\n",
|
1096 |
+
"tokenizer.save_pretrained(pt_save_directory)\n",
|
1097 |
+
"model.save_pretrained(pt_save_directory)"
|
1098 |
+
]
|
1099 |
+
},
|
1100 |
+
{
|
1101 |
+
"cell_type": "code",
|
1102 |
+
"execution_count": null,
|
1103 |
+
"id": "7df8097d-40f9-4524-8103-161dad3be122",
|
1104 |
+
"metadata": {},
|
1105 |
+
"outputs": [],
|
1106 |
+
"source": []
|
1107 |
+
}
|
1108 |
+
],
|
1109 |
+
"metadata": {
|
1110 |
+
"kernelspec": {
|
1111 |
+
"display_name": "Python 3 (ipykernel)",
|
1112 |
+
"language": "python",
|
1113 |
+
"name": "python3"
|
1114 |
+
},
|
1115 |
+
"language_info": {
|
1116 |
+
"codemirror_mode": {
|
1117 |
+
"name": "ipython",
|
1118 |
+
"version": 3
|
1119 |
+
},
|
1120 |
+
"file_extension": ".py",
|
1121 |
+
"mimetype": "text/x-python",
|
1122 |
+
"name": "python",
|
1123 |
+
"nbconvert_exporter": "python",
|
1124 |
+
"pygments_lexer": "ipython3",
|
1125 |
+
"version": "3.10.6"
|
1126 |
+
}
|
1127 |
+
},
|
1128 |
+
"nbformat": 4,
|
1129 |
+
"nbformat_minor": 5
|
1130 |
+
}
|
special_tokens_map.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<s>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": true,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "</s>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": true,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": "</s>",
|
17 |
+
"unk_token": {
|
18 |
+
"content": "<unk>",
|
19 |
+
"lstrip": false,
|
20 |
+
"normalized": true,
|
21 |
+
"rstrip": false,
|
22 |
+
"single_word": false
|
23 |
+
}
|
24 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
3 |
+
size 499723
|
tokenizer_config.json
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": true,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"bos_token": {
|
5 |
+
"__type": "AddedToken",
|
6 |
+
"content": "<s>",
|
7 |
+
"lstrip": false,
|
8 |
+
"normalized": true,
|
9 |
+
"rstrip": false,
|
10 |
+
"single_word": false
|
11 |
+
},
|
12 |
+
"clean_up_tokenization_spaces": false,
|
13 |
+
"eos_token": {
|
14 |
+
"__type": "AddedToken",
|
15 |
+
"content": "</s>",
|
16 |
+
"lstrip": false,
|
17 |
+
"normalized": true,
|
18 |
+
"rstrip": false,
|
19 |
+
"single_word": false
|
20 |
+
},
|
21 |
+
"model_max_length": 2048,
|
22 |
+
"pad_token": null,
|
23 |
+
"padding_side": "right",
|
24 |
+
"sp_model_kwargs": {},
|
25 |
+
"tokenizer_class": "LlamaTokenizer",
|
26 |
+
"unk_token": {
|
27 |
+
"__type": "AddedToken",
|
28 |
+
"content": "<unk>",
|
29 |
+
"lstrip": false,
|
30 |
+
"normalized": true,
|
31 |
+
"rstrip": false,
|
32 |
+
"single_word": false
|
33 |
+
}
|
34 |
+
}
|