Project-MONAI / Project-MONAI/tutorials

Pretrained model not compatible with downstream training

Open
#1,249 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Jupyter Notebook
Stars
2.5k
Forks
803
Avg merge
6d 22h
Merged PRs (30d)
3

Description

I am having some issues using a private pretrained encoder weights. When following the swin-unetr pretrain repo to train a pretrained encoder weights, the model can't be loaded for downstream segmentation training. Here is the error:

KeyError: 'module.patch_embed.proj.weight'

Is there anything that might be missing from the pretrain repo? Would it be possible to fix this issue without retraining? Thank you!!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the swin-unetr pretrain repo and the downstream segmentation training path, then trace how the private pretrained encoder weights are loaded when the KeyError for module.patch_embed.proj.weight occurs. Done means determining whether the pretrained weights can load for downstream segmentation without retraining, or documenting the missing compatibility requirement.

Written by the indexing model from the issue text.

Assessment

Tech stack
jupyter-notebook, pytorch
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.