search

ai4protein / Pro-FSFP

Pro-FSFP: Few-Shot Protein Fitness Prediction
GNU General Public License v3.0
59 stars 3 forks source link

NameError: name 'model_names' is not defined. Did you mean: 'model_name' #7

Open shangguan77 opened 2 months ago

shangguan77 commented 2 months ago

Hello, I'm having some problems. Error when running "python retrieve.py -m vectorize -md esm2" : NameError: name 'model_names' is not defined. Did you mean: 'model_name'?

The error occurred in fsfp/retrieval.py. def get_base_model(self): model_name = model_names[self.args.model] model = EsmForMaskedLM.from_pretrained(model_name, output_hidden_states=True) tokenizer = EsmTokenizer.from_pretrained(model_name) return model, tokenizer

I find that model_names is not imported, so an error is reported, but I do not know how to change the code, can you solve this problem, I would be very grateful if you can.

ai4protein commented 2 months ago

Thank you for figuring out this bug! We have updated the fsfp/retrieval.py. The code 'model_name = model_names[self.args.model]' should be 'model_name = config.model_dir[self.args.model]'.

shangguan77 commented 1 month ago

Thank you very much. This problem has been solved.However, a new problem has arisen. When running the code in run.sh, an error message appears: "main.py: error: unrecognized arguments: GEMME. Why is this happening?"