Issues

• [FINO] Guide gradient normalization cancels relative loss_weight multipliers

  facebookresearch/dinov3#393 · Filter by author minchoe828-gif opened on Oct 23, 2026

• [FINO] HPA TRAIN_SSL missing antibody labels (-1) reach the prototypical loss

  facebookresearch/dinov3#392 · Filter by author minchoe828-gif opened on Oct 23, 2026

• License question - score paper shooting targets?

  facebookresearch/dinov3#391 · Filter by author carkrueger opened on Oct 22, 2026

• Request for reconsideration of DINOv3 access — hans-liu-vla

  facebookresearch/dinov3#389 · Filter by author AIDD0214 opened on Oct 15, 2026

• Question regarding License Scope: Does Knowledge Distillation create a "Derivative Work"?

  facebookresearch/dinov3#388 · Filter by author m-cho-bunsekiya opened on Oct 13, 2026

• Access request rejected – appeal for academic use

  facebookresearch/dinov3#387 · Filter by author Peter-max-hub opened on Oct 9, 2026

• Inconsistency in weight decay description vs. implementation

  facebookresearch/dinov3#386 · Filter by author johnlyzhou opened on Oct 6, 2026

• License question: does the "Built with DINOv3" attribution apply to outputs generated with the model?

  facebookresearch/dinov3#384 · Filter by author dennis2vsteaks opened on Oct 4, 2026

• Does Section 1(v) apply to fictional weapons in a commercial video game?

  facebookresearch/dinov3#372 · Filter by author MrKiwiBirdman opened on Sep 26, 2026

• Potentialy bug in Open-vocab implementation

  facebookresearch/dinov3#355 · Filter by author blaz-r opened on Aug 13, 2026

• License clarification: on-site commercial demo without redistribution of DINOv3 weights/code

  facebookresearch/dinov3#348 · Filter by author hanasaan opened on Jun 16, 2026

• Quantitative Results about Video Classification

  facebookresearch/dinov3#339 · Filter by author J-am-ack opened on May 9, 2026

• Segmentation fault on RTX 5090 with CUDA 13 during repeated PyTorch CUDA forward passes

  facebookresearch/dinov3#338 · Filter by author Ol3x opened on May 4, 2026

• Trellis-2

  facebookresearch/dinov3#335 · Filter by author avdcdc opened on Mar 27, 2026
  1

• [Question] Satellite Resolution and Distribution Shifts

  facebookresearch/dinov3#333 · Filter by author moienr opened on Mar 26, 2026
  1

• Vit base pretraining config

  facebookresearch/dinov3#331 · Filter by author Flaick opened on Mar 22, 2026

• Using Imagenet mean/std for geobench classification

  facebookresearch/dinov3#321 · Filter by author raviolihui opened on Mar 3, 2026
  

• Linear Evaluation Result Incomplete Problem

  facebookresearch/dinov3#320 · Filter by author ywyz233 opened on Mar 1, 2026

• [ACTION NEEDED] Make DinoV3 available for users in Syria

  facebookresearch/dinov3#319 · Filter by author ayhmdalila opened on Feb 25, 2026
  1

• Extra dependencies are imported when loading models via torch.hub.load

  facebookresearch/dinov3#318 · Filter by author chenzhekl opened on Feb 23, 2026

• Release of ViT-S/16 distilled weights for the SAT-493M (Satellite) dataset

  facebookresearch/dinov3#317 · Filter by author isuchy opened on Feb 21, 2026

• covert pytorch to onnx

  facebookresearch/dinov3#312 · Filter by author newbecoder opened on Feb 10, 2026

• finetuning on custom dataset

  facebookresearch/dinov3#311 · Filter by author eshasadia opened on Feb 6, 2026

• Replicating geobench results

  facebookresearch/dinov3#310 · Filter by author raviolihui opened on Jan 29, 2026
  2

  

• pretrain on imagenet，The training results are inconsistent with the repo results

  facebookresearch/dinov3#309 · Filter by author Elijah-Yi opened on Jan 27, 2026
  6