Update README.md
Browse files
README.md
CHANGED
|
@@ -25,8 +25,9 @@ GAI-LLM/ko-llama2-toy is an auto-regressive language model based on the LLaMA2 t
|
|
| 25 |
|
| 26 |
**Training Dataset**
|
| 27 |
|
| 28 |
-
- We combined Open Korean Dateset.
|
| 29 |
-
-
|
|
|
|
| 30 |
|
| 31 |
# **Model Benchmark**
|
| 32 |
|
|
|
|
| 25 |
|
| 26 |
**Training Dataset**
|
| 27 |
|
| 28 |
+
- We combined Open Korean Dateset using mixed-strategy.
|
| 29 |
+
- Kopen-platypus + Everythinglm v2 + jojo0217/korean_rlhf_dataser + sentineg + hellaswag + copa
|
| 30 |
+
- We use A100 GPU 80GB * 8, when training.
|
| 31 |
|
| 32 |
# **Model Benchmark**
|
| 33 |
|