All issues

Search results

Only reporting the accuracy of the first gpu by eval_linear.py

Status: Open.

#289 In facebookresearch/dino;

· aitrics-chris opened on Aug 4, 2025

loss increase sharply after the first epoch

Status: Open.

#287 In facebookresearch/dino;

· chuzhidaojiaoshenme opened on Jun 18, 2025

ResNet backbone for DINO trainig

Status: Open.

#286 In facebookresearch/dino;

· williamsuryap opened on May 19, 2025

After fine-tuning, the model performs worse than the pre-trained model.

Status: Open.

#284 In facebookresearch/dino;

· xdwjc opened on Feb 26, 2025

Changing number of classes

Status: Open.

#282 In facebookresearch/dino;

· vikashg opened on Jul 31, 2024

Reproducing results on segmentation using training from scratch

Status: Open.

#281 In facebookresearch/dino;

· sonamsingh19 opened on Jul 22, 2024

Cannot reproduce KNN performance for vanilla ViT-S training

Status: Open.

#280 In facebookresearch/dino;

· wangh09 opened on Jun 28, 2024

Can I use dino model to match images with pixels? Just like with the clip model you can match pixels with text.

Status: Open.

#278 In facebookresearch/dino;

· dajiebi opened on May 21, 2024

How to convert to onnx

Status: Open.

#276 In facebookresearch/dino;

· Ochre-amber opened on May 2, 2024

Why Teacher network perform bettert than Student one during training?

Status: Open.

#274 In facebookresearch/dino;

· Hongbo-Z opened on Apr 9, 2024

Loss is not decresing

Status: Open.

#273 In facebookresearch/dino;

· risingClouds opened on Mar 7, 2024

Best features for image similarity

Status: Open.

#272 In facebookresearch/dino;

· fbliman opened on Mar 6, 2024