Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- README.md +2 -2
- proteinmoe_architecture.png +3 -0
.gitattributes
CHANGED
@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
inverse_folding.png filter=lfs diff=lfs merge=lfs -text
|
|
|
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
inverse_folding.png filter=lfs diff=lfs merge=lfs -text
|
37 |
+
proteinmoe_architecture.png filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
@@ -66,8 +66,8 @@ We fine-tuned a pretrained masked language model using MSA data by concatenating
|
|
66 |
| Sequence length | 2048 | 12800 | 12800 |
|
67 |
| Per Device Micro Batch Size | 1 | 1 | 1 |
|
68 |
| Precision | Mixed FP32-FP16 | Mixed FP32-FP16 | Mixed FP32-FP16 |
|
69 |
-
|
|
70 |
-
|
|
71 |
|
72 |
### Tokenization
|
73 |
|
|
|
66 |
| Sequence length | 2048 | 12800 | 12800 |
|
67 |
| Per Device Micro Batch Size | 1 | 1 | 1 |
|
68 |
| Precision | Mixed FP32-FP16 | Mixed FP32-FP16 | Mixed FP32-FP16 |
|
69 |
+
| LR | [5e-6,5e-5] | [1e-6, 1e-5] | 1e-5 |
|
70 |
+
| Num Tokens | 10 billion | 100 billion | 80 billion |
|
71 |
|
72 |
### Tokenization
|
73 |
|
proteinmoe_architecture.png
ADDED
![]() |
Git LFS Details
|