File size: 2,818 Bytes
3ad581d |
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 |
{
"cells": [
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"import torch\n",
"from torch import Tensor\n",
"\n",
"from self_rewarding_lm_pytorch import (\n",
" SelfRewardingTrainer,\n",
" create_mock_dataset\n",
")\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"sft fine-tuning: 100%|██████████| 9/9 [00:03<00:00, 2.91it/s]\n",
"generating dpo dataset with self-rewarding: 0it [00:00, ?it/s]"
]
}
],
"source": [
"from x_transformers import TransformerWrapper, Decoder\n",
"transformer = TransformerWrapper(\n",
" num_tokens = 256,\n",
" max_seq_len = 1024,\n",
" attn_layers = Decoder(\n",
" dim = 512,\n",
" depth = 1,\n",
" heads = 8\n",
" )\n",
")\n",
"\n",
"sft_dataset = create_mock_dataset(100, lambda: (torch.randint(0, 256, (256,)), torch.tensor(1))) # length, output(callable function) -> return class instance\n",
"prompt_dataset = create_mock_dataset(100, lambda: 'mock prompt')\n",
"\n",
"def decode_tokens(tokens: Tensor) -> str:\n",
" decode_token = lambda token: str(chr(max(32, token))) # chr(i) : return ASCII code correspoding to i\n",
" return ''.join(list(map(decode_token, tokens)))\n",
"\n",
"def encode_str(seq_str: str) -> Tensor:\n",
" return Tensor(list(map(ord, seq_str))) # ord('c') : return the ASCII code of 'c'\n",
"\n",
"trainer = SelfRewardingTrainer(\n",
" transformer,\n",
" finetune_configs = dict(\n",
" train_sft_dataset = sft_dataset,\n",
" self_reward_prompt_dataset = prompt_dataset,\n",
" dpo_num_train_steps = 1000\n",
" ),\n",
" tokenizer_decode = decode_tokens,\n",
" tokenizer_encode = encode_str,\n",
" accelerate_kwargs = dict(\n",
" cpu = True\n",
" )\n",
")\n",
"trainer(overwrite_checkpoints = True)\n",
"\n",
"\n",
"# checkpoints after each finetuning stage will be saved to ./checkpoints"
]
},
{
"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": "3.11.7"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
|