Update README.md
Browse files
README.md
CHANGED
|
@@ -11,6 +11,7 @@ pipeline_tag: translation
|
|
| 11 |
This is an experimental LLaMA2 7B lora created using the [VNTL-v2.5-1k](https://huggingface.co/datasets/lmg-anon/VNTL-v2.5-1k) dataset.
|
| 12 |
|
| 13 |
This is a update of version 0.3:
|
|
|
|
| 14 |
- 2 epoches -> 1 epoch (2 epoches seemed to increase eval loss)
|
| 15 |
- Added EOS after each translation pair.
|
| 16 |
|
|
|
|
| 11 |
This is an experimental LLaMA2 7B lora created using the [VNTL-v2.5-1k](https://huggingface.co/datasets/lmg-anon/VNTL-v2.5-1k) dataset.
|
| 12 |
|
| 13 |
This is a update of version 0.3:
|
| 14 |
+
- adamw_bnb_8bit -> adamw_8bit (this is the default in unsloth)
|
| 15 |
- 2 epoches -> 1 epoch (2 epoches seemed to increase eval loss)
|
| 16 |
- Added EOS after each translation pair.
|
| 17 |
|