Error when running `python inference.py ...`

[walidabualafia]

Hi all,

Thanks for putting this software out.

I am having an issue running inference.py when I build the conda env. Please see the error below. The error is thousands of lines long:

/home/wabuala/E3-CryoFold/conda-env/lib/python3.9/site-packages/torch/_utils.py:776: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will
 be the only storage class. This should only matter to you if you are using storages directly.  To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.stor
age()
  return self.fget.__get__(instance, owner)()
Traceback (most recent call last):
  File "/research/rgs01/home/clusterHome/wabuala/E3-CryoFold/inference.py", line 32, in load_model
    cryofold.load_state_dict(checkpoint)
  File "/home/wabuala/E3-CryoFold/conda-env/lib/python3.9/site-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for CryoFold:
        Missing key(s) in state_dict: "esm.embed_tokens.weight", "esm.layers.0.self_attn.k_proj.weight", "esm.layers.0.self_attn.k_proj.bias", "esm.layers.0.self_attn.v_proj.weight",
 "esm.layers.0.self_attn.v_proj.bias", "esm.layers.0.self_attn.q_proj.weight", "esm.layers.0.self_attn.q_proj.bias", "esm.layers.0.self_attn.out_proj.weight", "esm.layers.0.self_attn
.out_proj.bias", "esm.layers.0.self_attn.rot_emb.inv_freq", "esm.layers.0.self_attn_layer_norm.weight", "esm.layers.0.self_attn_layer_norm.bias", "esm.layers.0.fc1.weight", "esm.laye
rs.0.fc1.bias", "esm.layers.0.fc2.weight", "esm.layers.0.fc2.bias", "esm.layers.0.final_layer_norm.weight", "esm.layers.0.final_layer_norm.bias", "esm.layers.1.self_attn.k_proj.weigh
t", "esm.layers.1.self_attn.k_proj.bias", "esm.layers.1.self_attn.v_proj.weight", ...

Please let me know if I am missing anything, or if there's something I can do to get this running.

Thanks!
Walid

[wangjue1111]

Hi Walid,

I sincerely apologize for the delay in my response. Some users reported issues with the last version of E3-CryoFold, which led us to temporarily remove the pre-trained model while we addressed these concerns. I’m pleased to inform you that we have now updated both the code and the pre-trained model. You are welcome to use them, and I hope they prove helpful to you.

Best regards,
Jue Wang