Update README.md

README.md

@@ -46,7 +46,7 @@ You can compute structure predictions based on FASTA sequences using
 python predict.py -i <FASTA> -o <OUTPUT_DIRECTORY>
 ```
 
-The ProtT5 language model will be downloaded on first use and stored by default in the directory 'ProtT5-XL-U50'. You can change this directory with the `--t5_model` parameter.
+The ProtT5 protein language model used to generate sequence embeddings will be downloaded on first use (~7 GB) and stored by default in the directory 'ProtT5-XL-U50'. You can change this directory with the `--t5_model` parameter.
 By default, the script will produce PDB files and distance maps. You can disable outputs using the parameters `--no-pdb` and `--no-distance-maps` respectively.
 
 ## Mutation effect prediction