Commit
•
9d54d52
1
Parent(s):
7275aa5
Update README.md
Browse files
README.md
CHANGED
|
@@ -5,7 +5,8 @@ license: mit
|
|
| 5 |
# ESM-2 Full Finetune for Binding Sites
|
| 6 |
|
| 7 |
This model is a full finetune of ESM-2, to illustrate how full finetuning overfits and generalizes quite poorly compared to
|
| 8 |
-
LoRA and QLoRA finetuning. This model was finetuned on the 600K dataset.
|
|
|
|
| 9 |
|
| 10 |
```python
|
| 11 |
Train metrics:
|
|
|
|
| 5 |
# ESM-2 Full Finetune for Binding Sites
|
| 6 |
|
| 7 |
This model is a full finetune of ESM-2, to illustrate how full finetuning overfits and generalizes quite poorly compared to
|
| 8 |
+
LoRA and QLoRA finetuning. This model was finetuned on the 600K dataset. We also note that on the 24GB A10 GPU, the batch size
|
| 9 |
+
has to be significantly smaller.
|
| 10 |
|
| 11 |
```python
|
| 12 |
Train metrics:
|