Update README.md
Browse files
README.md
CHANGED
|
@@ -10,11 +10,11 @@ During the pre-training phase, we calculate the InfoNCE loss for each two modali
|
|
| 10 |
does.
|
| 11 |
|
| 12 |
## Model architecture
|
| 13 |
-
Protein sequence encoder
|
| 14 |
|
| 15 |
-
Protein structure encoder
|
| 16 |
|
| 17 |
-
Text encoder
|
| 18 |
|
| 19 |
## Obtain embeddings and calculate similarity score (please clone our repo first)
|
| 20 |
```
|
|
|
|
| 10 |
does.
|
| 11 |
|
| 12 |
## Model architecture
|
| 13 |
+
**Protein sequence encoder**: [esm2_t12_35M_UR50D](https://huggingface.co/facebook/esm2_t12_35M_UR50D)
|
| 14 |
|
| 15 |
+
**Protein structure encoder**: foldseek_t12_35M (identical architecture with esm2 except that the vocabulary only contains 3Di tokens)
|
| 16 |
|
| 17 |
+
**Text encoder**: [BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext](https://huggingface.co/microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext)
|
| 18 |
|
| 19 |
## Obtain embeddings and calculate similarity score (please clone our repo first)
|
| 20 |
```
|