Search results

SigLIP 2: Poor Text-to-Image Retrieval Accuracy for Fine-Grained Attribute Queries

Status: Open.

#186 In google-research/big_vision;

· zhengthomastang opened on Dec 5, 2025

Image Retrival with Cos Sim

Status: Open.

#184 In google-research/big_vision;

· Matagi1996 opened on Oct 14, 2025

input_pipeline_test missing

Status: Open.

#183 In google-research/big_vision;

· AlexGokan opened on Sep 29, 2025

The performance of NaFlex for VLMs

Status: Open.

#182 In google-research/big_vision;

· Lele-Xie opened on Sep 26, 2025

On the effectiveness of the NaFlex variant for high-resolution images

Status: Open.

#181 In google-research/big_vision;

· hm123450 opened on Sep 6, 2025

Why the loss is divided by n, not n^2?

Status: Open.

#179 In google-research/big_vision;

· charlesCXK opened on Jul 11, 2025

Have you provide the code for finetuning Siglip2 using all the loss (captioning, dense captioning, self-distillation, masked prediciton) as mentioned in the original paper?

Status: Open.

#178 In google-research/big_vision;

· Li-Xiaoguang opened on Jun 26, 2025

Installation

Status: Open.

#175 In google-research/big_vision;

· jiaxi98 opened on May 17, 2025

siglip or siglip2 image process

Status: Open.

#173 In google-research/big_vision;

· zxqing01 opened on Apr 26, 2025

if I want to use my image-text data to finetune the SigLIP2， where I can get the train code?

Status: Open.

#172 In google-research/big_vision;

· LeopardCatCat opened on Apr 19, 2025

Open-Vocabulary semantic segmentation

Status: Open.

#171 In google-research/big_vision;

· zxqing01 opened on Apr 12, 2025

The ratio of reduced memory with sigmoid loss

Status: Open.

#170 In google-research/big_vision;

· Sorr7maker opened on Mar 21, 2025