|
--- |
|
license: mit |
|
datasets: |
|
- heegyu/hh-rlhf-ko |
|
- maywell/ko_Ultrafeedback_binarized |
|
- MrBananaHuman/kor_ethical_question_answer |
|
- heegyu/PKU-SafeRLHF-ko |
|
language: |
|
- ko |
|
--- |
|
|
|
- Base Model: [42dot/42dot_LLM-SFT-1.3B](https://huggingface.co/42dot/42dot_LLM-SFT-1.3B) |
|
|
|
## Hyperparameters: |
|
- Batch: 128 |
|
- Learning Rate: 1e-5 -> 1e-6 (Linear Decay) |
|
- Optimizer: AdamW (beta1 = 0.9, beta2 = 0.999) |
|
- Epoch: 2 (main revisionμ 1 epoch) |
|
- [νμ΅ report](https://wandb.ai/heegyukim/reward_model/reports/ko-reward-model-1-3b-v0-1---Vmlldzo2MTk2MDY2) |
|
## Performance |
|
| Dataset | Accuracy (epoch=1) | |
|
|----------------------------|--------------------| |
|
| hh-rlhf-ko | 59.02 | |
|
| hh-rlhf-ko (helpful) | 64.72 | |
|
| hh-rlhf-ko (harmless) | 44.29 | |
|
| ko-skku-rlhf | 68.69 | |
|
| PKU-SafeRLHF-ko (safer) | 64.09 | |
|
| kor-ethical-qa | 99.8 | |
|
| ko-ultrafeedback-binarized | 74.96 | |
|
| Average | 64.71 | |
|
|
|
|
|
## Usage |
|
- κΈ°μ‘΄ 42dot SFT λͺ¨λΈμ λν ν
νλ¦Ώμ μ¬μ©. |
|
- μ¬μ©μμ λ°νλ `<user>:\n`λ‘ μμ |
|
- Botμ λ°νλ `<bot>:\n`μΌλ‘ μμ |
|
|
|
|
|
``` |
|
from transformers import pipeline |
|
|
|
pipe = pipeline("text-classification", model="heegyu/ko-reward-model-1.3b-v0.1") |
|
|
|
pipe("""<human>: |
|
κ΄νλ¬Έ κ΄μ₯ κ°λ λ°©λ² μλ €μ£Όμ€ μ μλμ? |
|
<bot>: |
|
μ«μ΄μ<|endoftext|>""") |
|
# [{'label': 'LABEL_0', 'score': 0.040634412318468094}] |
|
|
|
pipe("""<human>: |
|
κ΄νλ¬Έ κ΄μ₯ κ°λ λ°©λ² μλ €μ£Όμ€ μ μλμ? |
|
<bot>: |
|
κ΄νλ¬Έκ΄μ₯μΌλ‘ κ°λ λ°©λ²μ λ€μκ³Ό κ°μ΅λλ€: |
|
μ§νμ² 3νΈμ 경볡κΆμμμ νμ°¨ν ν 6λ² μΆκ΅¬λ‘ λμ μ λΆμ€μμ²μ¬, κ΄νλ¬Έ λ°©ν₯μΌλ‘ μ΄λν©λλ€. |
|
μ§νμ² 5νΈμ κ΄νλ¬Έμμμ νμ°¨ν ν ν΄μΉλ§λΉ μ°κ²°ν΅λ‘λ₯Ό μ΄μ©ν΄ 7λ² μΆκ΅¬λ‘ λμ κ΄μ₯μ² λ°©ν₯μΌλ‘ μ΄λν©λλ€. |
|
μ§νμ² 1νΈμ μμ²μμμ νμ°¨ν ν 3λ² μΆκ΅¬λ‘ λμ λμκΆμ μ§λ μ½λ¦¬μλ νΈν
λ°©ν₯μΌλ‘ μ΄λν©λλ€. |
|
λλ³΄λ‘ 2λΆ κ±°λ¦¬μ μλ μ’
κ°μμ μ΄μ©ν©λλ€. |
|
κ΄νλ¬Έκ΄μ₯μΌλ‘ κ°λ λ²μ€ λ
Έμ μ λ€μκ³Ό κ°μ΅λλ€: 272λ²γ401λ²γ406λ²γ704λ²γ7022λ² |
|
|
|
λμμ΄ λμ
¨μΌλ©΄ μ’κ² μ΅λλ€!<|endoftext|>""") |
|
# [{'label': 'LABEL_0', 'score': 0.2885928750038147}] |
|
|
|
|
|
pipe("""<human>: |
|
λ§μ½μ μ΄λμμ ꡬν μ μμ΄μ? |
|
|
|
<bot>: |
|
μ λ ΄νκ² κ΅¬ν μ μλ κ³³μ μλ΄ν΄λλ¦¬κ² μ΅λλ€. <|endoftext|>""") |
|
# [{'label': 'LABEL_0', 'score': 0.25814417004585266}] |
|
|
|
|
|
pipe("""<human>: |
|
λ§μ½μ μ΄λμμ ꡬν μ μμ΄μ? |
|
|
|
<bot>: |
|
λ§μ½μ μ€λ
, κ±΄κ° λ¬Έμ , λ²μ λ¬Έμ λ₯Ό μ΄λνμ¬ μ¬κ°ν μνμ±μ λ΄ν¬νκ³ μμ΅λλ€. <|endoftext|>""") |
|
# [{'label': 'LABEL_0', 'score': 0.8125637173652649}] |
|
|
|
``` |