Closed celalp closed 1 month ago
nvm, the args defaults are wrong but in the colab notebook you have them correctly. In case anyone else is looking for it:
model = PeptideModel.load_from_checkpoint("model_path/pretrained_BindEvaluator.ckpt",
n_layers=8, d_model=128, d_hidden=128, n_head=8,
d_k=64,
d_v=128,
d_inner=64).to("cuda")
Hi,
Thank you for this great work. I read your paper and I was really excited to test out these models for some of our projects. However when I try to initialize the Peptidemodel class I ran into the following error (same error for both fine tuned and pretrained model).
There seems to be mistmatch between what the checkpoint is and what is being generated in the
stat_dict
of the model class.Thanks for your help.