TypeError in encoding.py: replace() argument 2 must be str, not None

[SeongIk-Choi]

Hi,

I'm running the generate.py script using the provided gen_configs.py, and I encountered the following error:

./lib/python3.11/site-packages/esm/utils/encoding.py", line 53, in tokenize_sequence
sequence = sequence.replace(C.MASK_STR_SHORT, sequence_tokenizer.mask_token)
TypeError: replace() argument 2 must be str, not None

It seems like sequence_tokenizer.mask_token is None during execution.
Do you have any insight into how to resolve this issue? Perhaps I am missing something when initializing the tokenizer?

Additionally, I tried to modify several parameters as mentioned in the inference script:

PDB name

substrate_smiles, product_smiles

n_pocket_res = 16, n_protein_res = 25

num_pocket_design_t = 25

However, changing these does not seem to have a clear effect or causes further errors.
Is there any guideline or recommendation for setting these parameters during inference?

Thanks in advance for your help!

[Di-Zhu123]

I also meet this problem and solve it by:
pip install transformers==4.46.2 tokenizers==0.20.3

[gaodagaoda123-ui]

Thank you. The method above successfully solved the problem.