Views
No views yet
1# https://pytorch.org/get-started/previous-versions/
2#pip install torch==2.5.0 torchvision==0.20.0 torchaudio==2.5.0 --index-url https://download.pytorch.org/whl/cu121
3
4#pip install -U pip
5
6#pip install -U transformers
7#pip install -U tokenizers
8#pip install -U bitsandbytes
9#pip install -U accelerate
10#pip install -U datasets
11#pip install -U peft
12#pip install -U trl
13
14from transformers import (
15 AutoModelForCausalLM,
16 AutoTokenizer,
17 BitsAndBytesConfig,
18)
19from peft import PeftModel
20import torch
21from tqdm import tqdm
22import json
23
24# Hugging face token
25HF_TOKEN = ""
26
27# model
28model_id = "llm-jp/llm-jp-3-13b"
29adapter_id = "shigedon/llm-jp-3-13b-finetune-16-Dec-2024"
30
31# QLoRA config
32bnb_config = BitsAndBytesConfig(
33 load_in_4bit=True,
34 bnb_4bit_quant_type="nf4",
35 bnb_4bit_compute_dtype=torch.bfloat16,
36)
37
38# Load model
39model = AutoModelForCausalLM.from_pretrained(
40 model_id,
41 quantization_config=bnb_config,
42 device_map="auto",
43 token = HF_TOKEN
44)
45
46# Load tokenizer
47tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True, token = HF_TOKEN)
48
49# 元のモデルにLoRAのアダプタを統合。
50model = PeftModel.from_pretrained(model, adapter_id, token = HF_TOKEN)
51
52# データセットの読み込み。
53# omnicampusの開発環境では、左にタスクのjsonlをドラッグアンドドロップしてから実行。
54datasets = []
55with open("./elyza-tasks-100-TV_0.jsonl", "r") as f:
56 item = ""
57 for line in f:
58 line = line.strip()
59 item += line
60 if item.endswith("}"):
61 datasets.append(json.loads(item))
62 item = ""
63
64results = []
65for data in tqdm(datasets):
66
67 input = data["input"]
68 prompt = f"""### 指示
69 {input}
70 ### 回答
71 """
72
73 input_ids = tokenizer(prompt, return_tensors="pt",return_token_type_ids=False,).to(model.device)
74 outputs = model.generate(**input_ids, max_new_tokens=512, do_sample=False, repetition_penalty=1.2,)
75 output = tokenizer.decode(outputs[0][input_ids.input_ids.size(1):], skip_special_tokens=True)
76
77 results.append({"task_id": data["task_id"], "input": input, "output": output})
78
79
80## llmjp
81results = []
82for data in tqdm(datasets):
83
84 input = data["input"]
85
86 prompt = f"""### 指示
87 {input}
88 ### 回答
89 """
90
91 tokenized_input = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt").to(model.device)
92 attention_mask = torch.ones_like(tokenized_input)
93 with torch.no_grad():
94 outputs = model.generate(
95 tokenized_input,
96 attention_mask=attention_mask,
97 max_new_tokens=100,
98 do_sample=False,
99 repetition_penalty=1.2,
100 pad_token_id=tokenizer.eos_token_id
101 )[0]
102 output = tokenizer.decode(outputs[tokenized_input.size(1):], skip_special_tokens=True)
103
104 results.append({"task_id": data["task_id"], "input": input, "output": output})
105
106# jsonl
107import re
108jsonl_id = re.sub(".*/", "", adapter_id)
109with open(f"./{jsonl_id}-outputs.jsonl", 'w', encoding='utf-8') as f:
110 for result in results:
111 json.dump(result, f, ensure_ascii=False) # ensure_ascii=False for handling non-ASCII characters
112 f.write('\n')