Jihwan Eom

Enable resume

6

## This PR includes: ### Summary: - Enable resume for all tasks. (cls/det/seg) - Update MPA submodule to parse resume information in stage.py - Related MPA PR: https://github.com/openvinotoolkit/model_preparation_algorithm/pull/52 - Related...

Attention Map Visualization Script in "Vision Transformers Need Registers" Paper

2

Hi there, thanks for merging #282! I was wondering if you could release the script you used for visualizing the attention maps in the [VISION TRANSFORMERS NEED REGISTERS](https://arxiv.org/pdf/2309.16588.pdf) paper. It...

Clarifications on High-Resolution Adaptation

4

Hello, I'm want to clarify regarding the high-resolution adaptation described in the paper. As per section 4 and Appendix B.2, it's mentioned that the model was trained at a higher...

ViT inference does not work by FeatureVectorHook.

1

**Describe the bug** `otx train` throws `AttributeError: 'list' object has no attribute 'size'` for inference time by FeatureVectorHook. **Steps to Reproduce** ``` otx build --task classification --backbone mmcls.VisionTransformer cd otx-workspace-CLASSIFICATION...

**Describe the bug** https://github.com/openvinotoolkit/training_extensions/blob/develop/otx/algorithms/classification/adapters/mmcls/configurer.py#L632-L647 Current loss patching logic is dynamically and forcefully changing the loss type to either `CrossEntropy` or `IBLoss`. This code prevents the loss type from being updated...

Request for Release of Training & Evaluation Logs

3

Hello, Would it be possible for you to release the training and evaluation logs, similar to what was provided in the DINOv1 repository? It would be greatly appreciated. Thank you...