|
--- |
|
license: apache-2.0 |
|
--- |
|
## Model |
|
- base_model : yanolja/KoSOLAR-10.7B-v0.2 |
|
- training objective: freeze, instruction Tuning |
|
## Dataset |
|
공개 데이터 수집 |
|
- Deduplicating Training Data Makes Language Models Better 알고리즘 활용 |
|
- instruction version 1.4 |
|
|
|
## Code |
|
```python |
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
import torch |
|
|
|
model_name = "jjingyeom/freeze_KoSoLAR-10.7B-v0.2_1.4_dedup" |
|
model = AutoModelForCausalLM.from_pretrained( |
|
model_name, |
|
) |
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
``` |
