File size: 6,827 Bytes
912e66e 7ea57da 912e66e 7ea57da 912e66e 7ea57da 912e66e 7ea57da 912e66e 7ea57da 912e66e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 |
{
"cells": [
{
"cell_type": "code",
"execution_count": null,
"id": "3890292a-c99e-4367-955d-5883b93dba36",
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"!pip install -q peft transformers datasets huggingface_hub\n",
"!pip install flash-attn --no-build-isolation"
]
},
{
"cell_type": "code",
"execution_count": 20,
"id": "f1cc378f-afb6-441f-a4c6-2ec427b4cd4b",
"metadata": {},
"outputs": [],
"source": [
"from transformers import AutoModelForCausalLM, AutoTokenizer, default_data_collator, get_linear_schedule_with_warmup\n",
"from peft import get_peft_config, get_peft_model, PromptTuningInit, PromptTuningConfig, TaskType, PeftType\n",
"import torch\n",
"from datasets import load_dataset\n",
"import os\n",
"from torch.utils.data import DataLoader\n",
"from tqdm import tqdm\n",
"from huggingface_hub import notebook_login\n",
"from huggingface_hub import HfApi"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "e4ab50d7-a4c9-4246-acd8-8875b87fe0da",
"metadata": {},
"outputs": [],
"source": [
"notebook_login()"
]
},
{
"cell_type": "code",
"execution_count": 24,
"id": "8a1cb1f9-b89d-4cac-a595-44e1e0ef85b2",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"CommitInfo(commit_url='https://huggingface.co/Granther/prompt-tuned-phi3/commit/912e66e469c6dd381daaa1ee25f5284e17c9377a', commit_message='Upload prompt_tune_phi3.ipynb with huggingface_hub', commit_description='', oid='912e66e469c6dd381daaa1ee25f5284e17c9377a', pr_url=None, pr_revision=None, pr_num=None)"
]
},
"execution_count": 24,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"api = HfApi()\n",
"api.upload_file(path_or_fileobj='prompt_tune_phi3.ipynb',\n",
" path_in_repo='prompt_tune_phi3.ipynb',\n",
" repo_id='Granther/prompt-tuned-phi3',\n",
" repo_type='model'\n",
" )"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "6cad1e5c-038f-4e75-8c3f-8ce0a43713a4",
"metadata": {},
"outputs": [],
"source": [
"device = 'cuda'\n",
"\n",
"model_id = 'microsoft/Phi-3-mini-128k-instruct'\n",
"\n",
"peft_conf = PromptTuningConfig(\n",
" peft_type=PeftType.PROMPT_TUNING, # what kind of peft\n",
" task_type=TaskType.CAUSAL_LM, # config task\n",
" prompt_tuning_init=PromptTuningInit.TEXT, # Set to 'TEXT' to use prompt_tuning_init_text\n",
" num_virtual_tokens=8, # x times the number of hidden transformer layers\n",
" prompt_tuning_init_text=\"Classify if the tweet is a complaint or not:\",\n",
" tokenizer_name_or_path=model_id\n",
")\n",
"\n",
"dataset_name = \"twitter_complaints\"\n",
"checkpoint_name = f\"{dataset_name}_{model_id}_{peft_conf.peft_type}_{peft_conf.task_type}_v1.pt\".replace(\n",
" \"/\", \"_\"\n",
")\n",
"\n",
"text_col = 'Tweet text'\n",
"lab_col = 'text_label'\n",
"max_len = 64\n",
"lr = 3e-2\n",
"epochs = 50\n",
"batch_size = 8"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "6f677839-ef23-428a-bcfe-f596590804ca",
"metadata": {},
"outputs": [],
"source": [
"dataset = load_dataset('ought/raft', dataset_name, split='train')"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "c0c05613-7941-4959-ada9-49ed1093bec4",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['Unlabeled', 'complaint', 'no complaint']"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dataset.features['Label'].names\n",
"#>>> ['Unlabeled', 'complaint', 'no complaint']"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "14e2bc8b-b4e3-49c9-ae2b-5946e412caa5",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "d9e958c687dd493880d18d4f1621dad9",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Map (num_proc=10): 0%| | 0/50 [00:00<?, ? examples/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/plain": [
"'Unlabeled'"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Create lambda function\n",
"classes = [k.replace('_', ' ') for k in dataset.features['Label'].names]\n",
"dataset = dataset.map(\n",
" lambda x: {'text_label': [classes[label] for label in x['Label']]},\n",
" batched=True,\n",
" num_proc=10,\n",
")\n",
"\n",
"dataset[0]"
]
},
{
"cell_type": "code",
"execution_count": 16,
"id": "19f0865d-e490-4c9f-a5f4-e781ed270f47",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n"
]
},
{
"data": {
"text/plain": [
"[1, 853, 29880, 24025, 32000]"
]
},
"execution_count": 16,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"tokenizer = AutoTokenizer.from_pretrained(model_id)\n",
"\n",
"if tokenizer.pad_token_id == None:\n",
" tokenizer.pad_token_id = tokenizer.eos_token_id\n",
"\n",
"target_max_len = max([len(tokenizer(class_lab)['input_ids']) for class_lab in classes])\n",
"target_max_len # max length for tokenized labels\n",
"\n",
"tokenizer(classes[0])['input_ids'] \n",
"# Ids corresponding to the tokens in the sequence\n",
"# Attention mask is a binary tensor used in the transformer block to differentiate between padding tokens and meaningful ones"
]
},
{
"cell_type": "markdown",
"id": "e1a15150-4bd9-45a2-ba43-d0bbbd16e60d",
"metadata": {},
"source": [
"### Preprocess Function:\n",
"- Tokenize text and label\n",
"- Pad each example in the batch with tok.pad_token_id\n",
"- "
]
}
],
"metadata": {
"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": "3.10.13"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|