waiyanan commited on
Commit
5af5a82
·
verified ·
1 Parent(s): c47758e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +42 -5
README.md CHANGED
@@ -24,13 +24,17 @@ language:
24
  # 使用方法例
25
 
26
  1. 依存関係のインストール:
27
- ```
28
- bash !pip install transformers datasets unsloth
 
 
 
 
29
  ```
30
 
31
  2. モデルの読み込み:
32
 
33
- ```python:
34
  from unsloth import FastLanguageModel
35
  from peft import PeftModel
36
 
@@ -51,17 +55,50 @@ model = PeftModel.from_pretrained(model, lola_adapter_id, token=hf_token)
51
  ```
52
  3. 推論
53
 
54
- ```
 
55
  FastLanguageModel.for_inference(model)
56
 
57
  inputs = tokenizer([prompt], return_tensors = "pt").to(model.device)
58
 
59
  outputs = model.generate(**inputs, max_new_tokens = max_token, use_cache = True, temperature=0.5, top_p=0.9,do_sample=False, repetition_penalty=1.2)
60
 
61
- prediction = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
62
 
63
  ```
64
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
65
 
66
  This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
67
 
 
24
  # 使用方法例
25
 
26
  1. 依存関係のインストール:
27
+ ```bash
28
+ # 必要なライブラリをインストール
29
+ pip install unsloth
30
+ pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
31
+ pip install -U torch
32
+ pip install -U peft
33
  ```
34
 
35
  2. モデルの読み込み:
36
 
37
+ ```python
38
  from unsloth import FastLanguageModel
39
  from peft import PeftModel
40
 
 
55
  ```
56
  3. 推論
57
 
58
+
59
+ ```python
60
  FastLanguageModel.for_inference(model)
61
 
62
  inputs = tokenizer([prompt], return_tensors = "pt").to(model.device)
63
 
64
  outputs = model.generate(**inputs, max_new_tokens = max_token, use_cache = True, temperature=0.5, top_p=0.9,do_sample=False, repetition_penalty=1.2)
65
 
66
+ prediction = tokenizer.decode(outputs[0], skip_special_tokens=True)
67
 
68
  ```
69
 
70
+ 4. 評価データの読み込みと評価
71
+
72
+ ```python
73
+ import pandas as pd
74
+ from datasets import Dataset
75
+
76
+ datasets = []
77
+ file_path = <評価データ(jsonl)>のパス
78
+
79
+ df = pd.read_json(filename, orient='records', lines=True)
80
+
81
+ results = []
82
+
83
+ for r in iterrows():
84
+ imput = r["input"]
85
+ prompt = f"""### 指示\n{input} 簡潔に回答してください \n### 回答\n"""
86
+
87
+ inputs = tokenizer([prompt], return_tensors = "pt").to(model.device)
88
+
89
+ outputs = model.generate(**inputs, max_new_tokens = max_token, use_cache = True, temperature=0.5, top_p=0.9,do_sample=False, repetition_penalty=1.2)
90
+ prediction = tokenizer.decode(outputs[0], skip_special_tokens=True).split('\n### 回答')[-1]
91
+
92
+ results.append({"task_id": r["task_id"], "input": input, "output": prediction})
93
+
94
+ # 結果をjsonlで保存。
95
+ json_file_id = <保存ファイル名>
96
+ with open(f"/content/{json_file_id}_output.jsonl", 'w', encoding='utf-8') as f:
97
+ for result in results:
98
+ json.dump(result, f, ensure_ascii=False)
99
+ f.write('\n')
100
+
101
+ ```
102
 
103
  This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
104