Update README.md
Browse files
README.md
CHANGED
|
@@ -62,6 +62,8 @@ emb_1 = embedding_repr.last_hidden_state[1,:4]
|
|
| 62 |
|
| 63 |
**NOTE**: Please make sure to explicitly set the model to `float16` (`T5EncoderModel.from_pretrained('Rostlab/prot_t5_xl_half_uniref50-enc', torch_dtype=torch.float16)`) otherwise, the generated embeddings will be full precision.
|
| 64 |
|
|
|
|
|
|
|
| 65 |
### BibTeX entry and citation info
|
| 66 |
|
| 67 |
```bibtex
|
|
|
|
| 62 |
|
| 63 |
**NOTE**: Please make sure to explicitly set the model to `float16` (`T5EncoderModel.from_pretrained('Rostlab/prot_t5_xl_half_uniref50-enc', torch_dtype=torch.float16)`) otherwise, the generated embeddings will be full precision.
|
| 64 |
|
| 65 |
+
**NOTE**: Currently (06/2022) half-precision models cannot be used on CPU. If you want to use the encoder only version on CPU, you need to cast it to its full-precision version (`model=model.float()`).
|
| 66 |
+
|
| 67 |
### BibTeX entry and citation info
|
| 68 |
|
| 69 |
```bibtex
|