Conversation
stephprince
left a comment
stephprince
left a comment
There was a problem hiding this comment.
Added my comments! I was mainly reviewing for any changes that would break the core model training so I did not look in detail at the refinement_model module updates
There was a problem hiding this comment.
I don't think this file should be included
There was a problem hiding this comment.
I think this and debug_train_structure.sh and test_imports.py are leftovers from an AI debugging run so could be removed but if you are still using them feel free to leave
| self.pdb_dir = pdb_dir | ||
| self.pdb_prefix = pdb_prefix if pdb_prefix is not None else "fixed_" | ||
| self.pdb_ext = pdb_ext or ".pdb" |
There was a problem hiding this comment.
I think the changes in this file would break some of the current training code - could you adapt your training code to use these args and uncomment the code below
| # calculate the L1 divergence of the SAXS profiles | ||
| l1_loss = nn.L1Loss(reduction="sum") | ||
| loss = l1_loss(pred_saxs, true_saxs) | ||
| loss = l1_loss(pred_saxs, saxs) |
There was a problem hiding this comment.
could you revert these changes? I don't think saxs should be an input anymore in this function
| "openfold @ git+https://github.com/aqlaboratory/openfold.git@103d037", | ||
| "ml-collections==0.1.0", | ||
| "dm-tree==0.1.6", | ||
| "lightning==2.1.0", |
There was a problem hiding this comment.
| "lightning==2.1.0", |
This is already specified above
…enerate some test files.
2 participants
Hi Steph,
This is my random sample branch with the ensemble generation ability. Could you have a look at this? I checked and there is not much overlap or conflicts with the main branch.
Best
Feng Yu