File size: 4,227 Bytes
b72c61a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
---
license: mit
language:
  - ko
  - en
tags:
  - korean
  - reasoning
  - instruction-tuning
  - fine-tuning
  - llama
  - desspseek
  - distillation
  - sft
---

# 🧠 DeepSeek-R1-Distill-Llama-8B-Ko-Reasoning

> A large-scale Korean reasoning model fine-tuned from **deepseek-ai/DeepSeek-R1-Distill-Llama-8B**, designed to excel in logical and multi-hop reasoning tasks in Korean.

---

## 📌 Overview

**DeepSeek-R1-Distill-Llama-8B-Ko-Reasoning** is a fine-tuned version of [deepseek-ai/DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B), specifically optimized for **logical reasoning in Korean**. This model is part of a broader research initiative to explore:

- The **transition from multilingual reasoning LLMs** to **Korean-specialized reasoning models**
- The enhancement of **non-reasoning Korean language models** into **reasoning-capable variants**
- The development of open-access models that rival proprietary alternatives in complex reasoning tasks

This model was fine-tuned using a large-scale Korean-English instruction dataset containing diverse multi-hop questions, symbolic logic tasks, and human-crafted reasoning steps.

---

## 🧪 Benchmark Results

> - 📊 All benchmarks were measured using the **0-shot CoT (Chain-of-Thought)** method.
> - 📊 The **Score** represents either the **accuracy (%)** of correct answers or a rating on a **1-10 scale** from a judge model.
> - 📊 **LLM-as-a-judge** benchmarks were evaluated using **GPT-4o (2024-08-01-preview)**.

| **Benchmark**    | **Score**     |
|------------------|---------------|
| GPQA diamond     | 58.8          |
| GSM8K            | 55.3          |
| HAERAE           | 70.8          |
| KSM              | 71.2          |
| LogicKor         | 7.84          |
| Math500          | 81.4          |
| MT-Bench         | 7.44          |
| MT-Bench(Ko)     | 7.09          |

---

## 🧑‍💻 Usage

Install Transformers >= 4.50:

```bash
pip install -U transformers
```

Basic example:

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "DimensionSTP/DeepSeek-R1-Distill-Llama-8B-Ko-Reasoning"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)

prompt = "서울과 부산 중 어디가 더 커?"
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)

model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)
```

---

## 🧠 Base Model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B

The base model, [deepseek-ai/DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B), is a CoT LLM developed by the DeepSeek AI team, fine tuned from Llama 3.1 base.
For more technical details, refer to the [Deepseek R1 Technical Report](https://arxiv.org/pdf/2501.12948).

---

## 🧱 Model Architecture

| Property         | Value                  |
|------------------|------------------------|
| Architecture     | LlamaForCausalLM       |
| Parameters       | 8B                     |
| Context Length   | 131,072 tokens         |
| Tokenizer        | LLamaTokenizer (BPE)   |

---

## 📅 Release Date

**Mar 2025**  
This model was released in March 2025 as part of the **Ko-Reasoning Series**, which focuses on pushing the boundaries of open-source reasoning in Korean using modern LLMs.

---

## 📬 Contact

For questions, collaborations, or deployment inquiries, please contact:

- 🤖 Hugging Face: [https://huggingface.co/DimensionSTP](https://huggingface.co/DimensionSTP)
- ✉️ Email: [ddang8jh@gmail.com]

---

## 📦 Available Checkpoints

-`main`: Final stable version from the `last` branch
- ✅ All training artifacts available (tokenizer, config, model weights)