Update README.md
Browse files
README.md
CHANGED
|
@@ -1,24 +1,18 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
-
datasets:
|
| 4 |
-
- blenderwang/zinc-50M
|
| 5 |
---
|
| 6 |
|
| 7 |
# Mol ID
|
| 8 |
|
| 9 |
A transformer encoder model pretrained on 50M ZINC SMILES string using flash attention 2
|
| 10 |
|
| 11 |
-
With modern hardware and software stacks, it only took me 6~7 hours to pretrain this 17M model on 50M molecules for 5 epochs
|
| 12 |
-
|
| 13 |
Hardware:
|
| 14 |
-
-
|
| 15 |
-
- 1 RTX 3090 gpu
|
| 16 |
|
| 17 |
Software:
|
| 18 |
- flash attention 2
|
| 19 |
- lightning for mixed precision (bf16-mixed)
|
| 20 |
- wandb for logging
|
| 21 |
-
- [report](https://api.wandb.ai/links/blenderwang/5qou429x)
|
| 22 |
- huggingface
|
| 23 |
- tokenizers
|
| 24 |
- datasets
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
# Mol ID
|
| 6 |
|
| 7 |
A transformer encoder model pretrained on 50M ZINC SMILES string using flash attention 2
|
| 8 |
|
|
|
|
|
|
|
| 9 |
Hardware:
|
| 10 |
+
- gpu that support flash attention 2 and bf16
|
|
|
|
| 11 |
|
| 12 |
Software:
|
| 13 |
- flash attention 2
|
| 14 |
- lightning for mixed precision (bf16-mixed)
|
| 15 |
- wandb for logging
|
|
|
|
| 16 |
- huggingface
|
| 17 |
- tokenizers
|
| 18 |
- datasets
|