Fix/vision only forward - #785
Merged
Merged
Conversation
…ed configs Encoder3D used `i > num_single_blocks`, so the default of 2 gave three single-conv blocks. Now `>=`. Default max_channels drops from 256 to 128: the 256-channel blocks sat on 8^3 maps and held most of the parameters without adding anything the earlier stages hadn't already encoded. Both changes alter what a saved config means, so CNN3D configs now carry arch_version (2). CNN3D.upgrade_config maps version-1 configs (no key) onto the current constructor by adding one to num_single_blocks and pinning max_channels=256, so v4a-v4c checkpoints rebuild the exact architecture they trained with. Applied in CNN3D.load and in ArboristVisionMergeDetector.load for the CNN3D backbone. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Vision-only counterpart to ArboristVisionMergeDetector: classifies merge sites from 3D image patches with no graph input, using the CNN3D or ViT3D backbone with its built-in output head. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When modality="image", ThreadedDataLoader yields a plain tensor (not a dict), so x["img"] raised a TypeError. Now accepts both. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…e cap - _num_neg_candidates: when random_nonmerge_site_prob > 0, return max(n_nonmerge_sites, n_target) so indices above the annotated pool fall through to get_random_nonmerge_site() in get_site() - _list_indices: apply max_brain_examples cap by subsampling both classes proportionally; positives are re-sampled each epoch so all merge sites get coverage over training - create_dataset_collection: accept and forward max_brain_examples; cap is applied to train only (rebalance_classes=True) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.