lightly-ai / lightly-ai/lightly-train

[QUESTION] distill dinov3 to repvit

Open
#628 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

question
Dominant language
Python
Stars
1.7k
Forks
116
Avg merge
2d 21h
Merged PRs (30d)
6

Description

Hello, I'm trying to distill Dinov3 into a Repvit model using lightly-train. Referring to your [documentation](https://docs.lightly.ai/train/stable/pretrain_distill/models/custom_models.html#custom-models) on how to distill a custom model, I found that the `forward_features` function might return multiple layers of features, and these features might have inconsistent resolutions, making them impossible to concatenate into a single tensor. In this case, how can I supervise the multi-layer features?

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 custom-model distillation documentation linked in the issue and inspect how the model's forward_features output is handled. Compare the DINOv3 and RepViT feature outputs, then document whether multi-layer features with different resolutions are supported and how supervision should be configured.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
computer-vision, machine-learning
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.