Noelia Ferruz
commited on
Commit
·
006ad59
1
Parent(s):
d4c5995
Update README.md
Browse files
README.md
CHANGED
|
@@ -53,11 +53,7 @@ python run_clm.py --model_name_or_path nferruz/ProtGPT2 --train_file training.tx
|
|
| 53 |
The HuggingFace script run_clm.py can be found here: https://github.com/huggingface/transformers/blob/master/examples/pytorch/language-modeling/run_clm.py
|
| 54 |
|
| 55 |
### **How to select the best sequences**
|
| 56 |
-
We've observed that perplexity values correlate with AlphaFold2's plddt. This plot shows perplexity vs. pldtt values for each of the 10,000 sequences in the ProtGPT2-generated dataset
|
| 57 |
-
<div align="center">
|
| 58 |
-
<img src="https://huggingface.co/nferruz/ProtGPT2/blob/main/ppl-plddt.png" width="45%" />
|
| 59 |
-
</div>
|
| 60 |
-
|
| 61 |
|
| 62 |
We recommend to compute perplexity for each sequence with the HuggingFace evaluate method `perplexity`:
|
| 63 |
|
|
|
|
| 53 |
The HuggingFace script run_clm.py can be found here: https://github.com/huggingface/transformers/blob/master/examples/pytorch/language-modeling/run_clm.py
|
| 54 |
|
| 55 |
### **How to select the best sequences**
|
| 56 |
+
We've observed that perplexity values correlate with AlphaFold2's plddt. This plot shows perplexity vs. pldtt values for each of the 10,000 sequences in the ProtGPT2-generated dataset (see https://huggingface.co/nferruz/ProtGPT2/blob/main/ppl-plddt.png)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
|
| 58 |
We recommend to compute perplexity for each sequence with the HuggingFace evaluate method `perplexity`:
|
| 59 |
|