Hugging Face's logo

Namazu11
/
llm-jp-3-13b-sft-dpo2

Transformers
Safetensors
Japanese
text-generation-inference
unsloth
llama
trl
Inference Endpoints
Model card Files Community
llm-jp-3-13b-sft-dpo2
File size: 3,374 Bytes 
93a1412
173aefc
 
6ef2e6e
173aefc
 
 
 
 
 
 
c826639
 
 
 
 
 
93a1412
 
c826639
93a1412
c826639
 
 
c3b7561
93a1412
c826639
93a1412
6ef2e6e
 
 
 
aa0e4e4
 
 
 
 
 
 
 
 
 
d932409
aa0e4e4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
---
license: apache-2.0
datasets:
- ichikara-instruction
- elyza/ELYZA-tasks-100
- weblab-GENIAC/aya-ja-evol-instruct-calm3-dpo-masked
- llm-jp/hh-rlhf-12k-ja
language:
- ja
base_model:
- llm-jp/llm-jp-3-13b
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
---

# Uploaded  model

- **Developed by:** Namazu11
- **License:** apache-2.0
- **Finetuned from model :** llm-jp/llm-jp-3-13b
- **Dataset(タグ付け以外) :** ichikara-instruction(<a href="https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/">HPリンク</a>)

This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.

[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)


■このモデルは東京大学リスキリング講座「大規模言語モデル2024」の最終課題（コンペ）のためのものです。
「ELYZA-tasks-100-TV」というデータセットが配布され、精度を競います。

# Sample Use
```python
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    BitsAndBytesConfig,
)

HF_TOKEN = "Your Hugging Face Token"
base_model_id = "llm-jp/llm-jp-3-13b" 
adapter_id = "Namazu11/llm-jp-3-13b-sft-dpo2"

# QLoRA config
bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16,
)

# Load model
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    quantization_config=bnb_config,
    device_map="auto",
    token = HF_TOKEN
)

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True, token = HF_TOKEN)

# 元のモデルにLoRAのアダプタを統合。
model = PeftModel.from_pretrained(model, adapter_id, token = HF_TOKEN)

# データセットの読み込み。
datasets = []
with open("./elyza-tasks-100-TV_0.jsonl", "r") as f:
    item = ""
    for line in f:
      line = line.strip()
      item += line
      if item.endswith("}"):
        datasets.append(json.loads(item))
        item = ""

# 推論（llmjp）
results = []
for data in tqdm(datasets):

  input = data["input"]

  prompt = f"""### 指示
  {input}
  ### 回答
  """

  tokenized_input = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt").to(model.device)
  attention_mask = torch.ones_like(tokenized_input)
  with torch.no_grad():
      outputs = model.generate(
          tokenized_input,
          attention_mask=attention_mask,
          max_new_tokens=100,
          do_sample=False,
          repetition_penalty=1.2,
          pad_token_id=tokenizer.eos_token_id
      )[0]
  output = tokenizer.decode(outputs[tokenized_input.size(1):], skip_special_tokens=True)

  results.append({"task_id": data["task_id"], "input": input, "output": output})

# 出力結果のjsolファイル生成
import re
jsonl_id = re.sub(".*/", "", adapter_id)
with open(f"./{jsonl_id}-outputs.jsonl", 'w', encoding='utf-8') as f:
    for result in results:
        json.dump(result, f, ensure_ascii=False)  # ensure_ascii=False for handling non-ASCII characters
        f.write('\n')