File size: 1,706 Bytes
5f8220c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language: 
- en
- zh
- ja
base_model: Qwen/Qwen3-14B
tags:
- interactive-refinement
- multi-round-reasoning
- qwen3
- text-generation
license: apache-2.0
pipeline_tag: text-generation
---

# Interactive Refinement Qwen3-14B

このモデルは、Qwen3-14BをベースとしたInteractive Refinement(対話的改善)システムです。

## モデル概要

- **ベースモデル**: Qwen/Qwen3-14B
- **改善ラウンド数**: 3
- **1ラウンド最大トークン**: 600
- **量子化**: なし

## Interactive Refinementとは

Interactive Refinementは、同一の質問に対して複数ラウンドの推論を行い、各ラウンドで前回の回答を評価・改善する手法です。

### 主な特徴

1. **多段階推論**: 3回のラウンドで段階的に回答を改善
2. **思考過程の可視化**: `<think>`タグで思考過程を明示
3. **自己評価と改善**: 各ラウンドで前回の回答を評価し改善
4. **品質向上**: ベースライン単発生成より高品質な回答を生成

## 使用方法

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

# モデルとトークナイザーの読み込み
tokenizer = AutoTokenizer.from_pretrained("hiroshij/interactive-refinement-qwen3-14B")
model = AutoModelForCausalLM.from_pretrained("hiroshij/interactive-refinement-qwen3-14B")

# Interactive Refinement実行(実装は別途必要)
# question = "あなたの質問"
# response = interactive_refinement_generate(model, tokenizer, question)
```

## ライセンス

Apache 2.0

## 作成者

Interactive Refinement implementation based on Qwen3-14B

---

*このモデルは研究・教育目的で作成されました。*