babbler-1900 model doesnt work!

[hkoohy]

Hi, thank you for this nice and helpful work.
The example code doesnt work with any other options for pretrained model except 'bert-base'! I am especially interested in using 'babbler-1000' but I get an error saying 'cannot unpack non-iterable NoneType object'!

'''
from tape import ProteinBertModel, TAPETokenizer
model = ProteinBertModel.from_pretrained('babbler-1900') ##('bert-base')
tokenizer = TAPETokenizer(vocab='unirep') # iupac is the vocab for TAPE models, use unirep for the UniRep model

Pfam Family: Hexapep, Clan: CL0536

sequence = 'GCTVEDRCLIGMGAILLNGCVIGSGSLVAAGALITQ'
token_ids = torch.tensor([tokenizer.encode(sequence)])
output = model(token_ids)
sequence_output = output[0]
pooled_output = output[1]
'''

The error I get is:

Model name 'babbler-1900' was not found in model name list (bert-base). We assumed 'babbler-1900' was a path or url but couldn't find any file associated to this path or url.

TypeError Traceback (most recent call last)
Input In [25], in
1 from tape import ProteinBertModel, TAPETokenizer
----> 2 model = ProteinBertModel.from_pretrained('babbler-1900') ##('bert-base')
3 tokenizer = TAPETokenizer(vocab='unirep') # iupac is the vocab for TAPE models, use unirep for the UniRep model
5 # Pfam Family: Hexapep, Clan: CL0536

File ~/opt/anaconda3/envs/tape_env/lib/python3.8/site-packages/tape/models/modeling_utils.py:478, in ProteinModel.from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
476 # Load config
477 if config is None:
--> 478 config, model_kwargs = cls.config_class.from_pretrained(
479 pretrained_model_name_or_path, *model_args,
480 cache_dir=cache_dir, return_unused_kwargs=True,
481 # force_download=force_download,
482 # resume_download=resume_download,
483 **kwargs
484 )
485 else:
486 model_kwargs = kwargs

TypeError: cannot unpack non-iterable NoneType object

[franzigeiger]

You have to use UniRepModel instead of ProteinBertModel