Update model card for DPO v4
Browse files
README.md
CHANGED
|
@@ -1,3 +1,178 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- ko
|
| 5 |
+
- en
|
| 6 |
+
library_name: transformers
|
| 7 |
+
tags:
|
| 8 |
+
- finance
|
| 9 |
+
- korean
|
| 10 |
+
- stock-analysis
|
| 11 |
+
- reasoning
|
| 12 |
+
- dpo
|
| 13 |
+
base_model: Qwen/Qwen2.5-7B-Instruct
|
| 14 |
+
pipeline_tag: text-generation
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
# VELA (Vector-Encoded Learning Agent)
|
| 18 |
+
|
| 19 |
+
**ํ๊ตญ ์ฃผ์์์ฅ ์ ๋ฌธ AI ์ ๋๋ฆฌ์คํธ**
|
| 20 |
+
|
| 21 |
+
VELA๋ ํ๊ตญ ์ฃผ์์์ฅ ๋ด์ค ๋ถ์ ๋ฐ ํฌ์ ๋ฆฌ์์น๋ฅผ ์ํด ํนํ๋ 7B ํ๋ผ๋ฏธํฐ ์ธ์ด ๋ชจ๋ธ์
๋๋ค.
|
| 22 |
+
|
| 23 |
+
## Model Details
|
| 24 |
+
|
| 25 |
+
| ํญ๋ชฉ | ๋ด์ฉ |
|
| 26 |
+
|------|------|
|
| 27 |
+
| **Base Model** | Qwen/Qwen2.5-7B-Instruct |
|
| 28 |
+
| **Training Stage** | SFT + DPO v4 |
|
| 29 |
+
| **Parameters** | 7.6B |
|
| 30 |
+
| **Context Length** | 8,192 tokens |
|
| 31 |
+
| **Precision** | BFloat16 |
|
| 32 |
+
| **License** | Apache 2.0 |
|
| 33 |
+
|
| 34 |
+
## Training Pipeline
|
| 35 |
+
|
| 36 |
+
```
|
| 37 |
+
Qwen2.5-7B-Instruct
|
| 38 |
+
โ
|
| 39 |
+
SFT (930K samples)
|
| 40 |
+
- ํ๊ตญ ์ฃผ์ ๋ด์ค ๋ถ์
|
| 41 |
+
- ๋ฆฌ์์น ๋ฆฌํฌํธ ์์ฑ
|
| 42 |
+
- Reasoning Trace ํ์ต
|
| 43 |
+
โ
|
| 44 |
+
DPO v4 (7,681 pairs)
|
| 45 |
+
- ์ค๊ตญ์ด/์์ด leak ๊ต์
|
| 46 |
+
- ํ๊ตญ์ด ์ถ๋ ฅ ๊ฐํ
|
| 47 |
+
- ํ์ ์ค์ ํฅ์
|
| 48 |
+
โ
|
| 49 |
+
VELA v1.0
|
| 50 |
+
```
|
| 51 |
+
|
| 52 |
+
## Capabilities
|
| 53 |
+
|
| 54 |
+
- **๋ด์ค ์ํฅ ๋ถ์**: ์ฃผ์ ๊ด๋ จ ๋ด์ค์ ์์ฅ ์ํฅ๋ ์์ธก
|
| 55 |
+
- **๋ฆฌ์์น ๋ฆฌํฌํธ ์์ฑ**: ๊ตฌ์กฐํ๋ ํฌ์ ๋ถ์ ๋ณด๊ณ ์ ์์ฑ
|
| 56 |
+
- **Reasoning Trace**: ๋จ๊ณ๋ณ ๋ถ์ ์ฌ๊ณ ๊ณผ์ ์์ฑ
|
| 57 |
+
- **๋ค์ค ์์ค ์ข
ํฉ**: ๋ด์ค, ์์ธ, ์๊ธ ๋ฐ์ดํฐ ํตํฉ ๋ถ์
|
| 58 |
+
|
| 59 |
+
## Usage
|
| 60 |
+
|
| 61 |
+
### Transformers
|
| 62 |
+
|
| 63 |
+
```python
|
| 64 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 65 |
+
import torch
|
| 66 |
+
|
| 67 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 68 |
+
"intrect/vela",
|
| 69 |
+
torch_dtype=torch.bfloat16,
|
| 70 |
+
device_map="auto"
|
| 71 |
+
)
|
| 72 |
+
tokenizer = AutoTokenizer.from_pretrained("intrect/vela")
|
| 73 |
+
|
| 74 |
+
messages = [
|
| 75 |
+
{"role": "system", "content": "๋น์ ์ ํ๊ตญ ์ฃผ์ ์ ๋ฌธ ์ ๋๋ฆฌ์คํธ์
๋๋ค."},
|
| 76 |
+
{"role": "user", "content": "์ผ์ฑ์ ์ HBM ์ฌ์
์ ๋ง์ ๋ถ์ํด์ฃผ์ธ์."}
|
| 77 |
+
]
|
| 78 |
+
|
| 79 |
+
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
| 80 |
+
inputs = tokenizer(text, return_tensors="pt").to(model.device)
|
| 81 |
+
|
| 82 |
+
outputs = model.generate(
|
| 83 |
+
**inputs,
|
| 84 |
+
max_new_tokens=1024,
|
| 85 |
+
temperature=0.7,
|
| 86 |
+
do_sample=True
|
| 87 |
+
)
|
| 88 |
+
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
| 89 |
+
```
|
| 90 |
+
|
| 91 |
+
### vLLM (Recommended for Production)
|
| 92 |
+
|
| 93 |
+
```python
|
| 94 |
+
from vllm import LLM, SamplingParams
|
| 95 |
+
|
| 96 |
+
llm = LLM(model="intrect/vela", dtype="bfloat16")
|
| 97 |
+
params = SamplingParams(temperature=0.7, max_tokens=1024)
|
| 98 |
+
|
| 99 |
+
prompts = ["์ผ์ฑ์ ์ HBM ์์ฅ ์ ๋ง์ ๋ถ์ํด์ฃผ์ธ์."]
|
| 100 |
+
outputs = llm.generate(prompts, params)
|
| 101 |
+
```
|
| 102 |
+
|
| 103 |
+
### MLX (Apple Silicon)
|
| 104 |
+
|
| 105 |
+
MLX ๋ณํ ๋ชจ๋ธ์ ๋ณ๋ ์ ์ฅ์์์ ์ ๊ณต ์์ ์
๋๋ค.
|
| 106 |
+
|
| 107 |
+
## Output Format
|
| 108 |
+
|
| 109 |
+
VELA๋ ๋ค์๊ณผ ๊ฐ์ ๊ตฌ์กฐํ๋ ์ถ๋ ฅ์ ์์ฑํฉ๋๋ค:
|
| 110 |
+
|
| 111 |
+
```markdown
|
| 112 |
+
## Executive Summary
|
| 113 |
+
[2-3๋ฌธ์ฅ ํต์ฌ ์์ฝ]
|
| 114 |
+
|
| 115 |
+
## Key Metrics
|
| 116 |
+
| ์งํ | ์์น |
|
| 117 |
+
|------|------|
|
| 118 |
+
| ํ์ฌ๊ฐ | โฉXX,XXX |
|
| 119 |
+
| PER | XX.X |
|
| 120 |
+
| ... | ... |
|
| 121 |
+
|
| 122 |
+
## ์์ฅ ๋ํฅ ๋ถ์
|
| 123 |
+
[์์ธ ๋ถ์]
|
| 124 |
+
|
| 125 |
+
## ๋ฆฌ์คํฌ ์์ธ
|
| 126 |
+
- ๋ฆฌ์คํฌ 1
|
| 127 |
+
- ๋ฆฌ์คํฌ 2
|
| 128 |
+
|
| 129 |
+
## ํฌ์ ์๊ฒฌ
|
| 130 |
+
[์ข
ํฉ ์๊ฒฌ]
|
| 131 |
+
```
|
| 132 |
+
|
| 133 |
+
## Training Data
|
| 134 |
+
|
| 135 |
+
| ๋ฐ์ดํฐ์
| ์ํ ์ | ์ฉ๋ |
|
| 136 |
+
|----------|---------|------|
|
| 137 |
+
| ํ๊ตญ ์ฃผ์ ๋ด์ค | 412K | SFT ๊ธฐ๋ฐ ๋ฐ์ดํฐ |
|
| 138 |
+
| ๋ฆฌ์์น ๋ฆฌํฌํธ | 50K | ๋ถ์ ํ์ ํ์ต |
|
| 139 |
+
| Reasoning Traces | 5K | ์ฌ๊ณ ๊ณผ์ ํ์ต |
|
| 140 |
+
| DPO Pairs | 7.7K | ์ ํธ๋ ์ ๋ ฌ |
|
| 141 |
+
|
| 142 |
+
## DPO v4 Improvements
|
| 143 |
+
|
| 144 |
+
DPO v4๋ ๋ค์ ๋ฌธ์ ๋ค์ ํด๊ฒฐํฉ๋๋ค:
|
| 145 |
+
|
| 146 |
+
- โ
**์ค๊ตญ์ด leak ์ ๊ฑฐ**: ์ค๊ตญ์ด ๋ฌธ์ ์ถ๋ ฅ ๋ฐฉ์ง
|
| 147 |
+
- โ
**์์ด leak ๊ฐ์**: ๋ถํ์ํ ์์ด ์ฌ์ฉ ์ต์ํ
|
| 148 |
+
- โ
**ํ์ ์ค์**: ์ง์ ๋ ์ถ๋ ฅ ํ์ ์๊ฒฉ ์ค์
|
| 149 |
+
- โ
**ํ๊ตญ์ด ํ์ง**: ์์ฐ์ค๋ฌ์ด ํ๊ตญ์ด ํํ
|
| 150 |
+
|
| 151 |
+
## Limitations
|
| 152 |
+
|
| 153 |
+
- ์ค์๊ฐ ์์ธ ๋ฐ์ดํฐ ์ ๊ทผ ๋ถ๊ฐ (์ธ๋ถ API ํ์)
|
| 154 |
+
- ํฌ์ ์กฐ์ธ์ด ์๋ ์ ๋ณด ์ ๊ณต ๋ชฉ์
|
| 155 |
+
- 8K ์ปจํ
์คํธ ์ ํ์ผ๋ก ๊ธด ๋ฌธ์ ์ฒ๋ฆฌ ํ๊ณ
|
| 156 |
+
|
| 157 |
+
## Citation
|
| 158 |
+
|
| 159 |
+
```bibtex
|
| 160 |
+
@misc{vela2025,
|
| 161 |
+
title={VELA: Vector-Encoded Learning Agent for Korean Stock Analysis},
|
| 162 |
+
author={intrect},
|
| 163 |
+
year={2025},
|
| 164 |
+
publisher={Hugging Face},
|
| 165 |
+
url={https://huggingface.co/intrect/vela}
|
| 166 |
+
}
|
| 167 |
+
```
|
| 168 |
+
|
| 169 |
+
## Version History
|
| 170 |
+
|
| 171 |
+
| ๋ฒ์ | ๋ ์ง | ๋ณ๊ฒฝ์ฌํญ |
|
| 172 |
+
|------|------|----------|
|
| 173 |
+
| v1.0 (DPO v4) | 2025-01-28 | DPO v4 ๋ณํฉ, ์ค๊ตญ์ด/์์ด leak ํด๊ฒฐ |
|
| 174 |
+
| v0.9 (SFT) | 2025-01-15 | SFT ๋ฒ ์ด์ค ๋ชจ๋ธ ๊ณต๊ฐ |
|
| 175 |
+
|
| 176 |
+
---
|
| 177 |
+
|
| 178 |
+
**Disclaimer**: ์ด ๋ชจ๋ธ์ ์ถ๋ ฅ์ ํฌ์ ์กฐ์ธ์ด ์๋๋๋ค. ๋ชจ๋ ํฌ์ ๊ฒฐ์ ์ ๋ณธ์ธ์ ํ๋จ๊ณผ ์ฑ
์ ํ์ ์ด๋ฃจ์ด์ ธ์ผ ํฉ๋๋ค.
|