Issues

• momentseeker数据不对

  TIGER-AI-Lab/VLM2Vec#212 · Filter by author liuzhiyong01 opened on Sep 8, 2026

• video.yaml配置不对

  TIGER-AI-Lab/VLM2Vec#211 · Filter by author liuzhiyong01 opened on Sep 8, 2026

• memory.yaml配置文件

  TIGER-AI-Lab/VLM2Vec#210 · Filter by author liuzhiyong01 opened on Sep 8, 2026

• v3 training data

  TIGER-AI-Lab/VLM2Vec#203 · Filter by author jaycheng6076 opened on May 17, 2026

• 评估支持使用vllm框架部署的模型吗

  TIGER-AI-Lab/VLM2Vec#200 · Filter by author sanshi9523 opened on Apr 3, 2026

• VD-Vidore-V2 evaluation metrics do not match the paper.

  TIGER-AI-Lab/VLM2Vec#194 · Filter by author Lebron-James777 opened on Mar 11, 2026
  2

• How to support eval Qwen2.5/3-omni

  TIGER-AI-Lab/VLM2Vec#193 · Filter by author jianghuyihei opened on Mar 5, 2026

• one gpu oom

  TIGER-AI-Lab/VLM2Vec#192 · Filter by author FightingFighting opened on Feb 11, 2026

• Does this code support training baseline models (GME, LamRA) on private datasets?

  TIGER-AI-Lab/VLM2Vec#191 · Filter by author xxzhang99 opened on Jan 19, 2026

• how to eval a openapi model

  TIGER-AI-Lab/VLM2Vec#189 · Filter by author simplew2011 opened on Jan 16, 2026

• esg_reports_human_labeled_v2: discrepancy between test corpus size and the number of uploaded images

  TIGER-AI-Lab/VLM2Vec#188 · Filter by author sahel-sh opened on Jan 16, 2026
  1

• why not directly use the original Qwen implementation on huggingface?

  Replied
  TIGER-AI-Lab/VLM2Vec#183 · Filter by author FightingFighting opened on Dec 20, 2025
  3

• Backward is missing in the training_step method of the GradCacheLateProcessTrainer

  TIGER-AI-Lab/VLM2Vec#181 · Filter by author kalik-murbanov opened on Dec 10, 2025

• Question on CLIP finetuning in VLM2Vec-v1

  TIGER-AI-Lab/VLM2Vec#178 · Filter by author jaeseokbyun opened on Nov 26, 2025

• gme-Qwen2-VL-2B-Instruct eval error

  bugevaluation/benchmark
  TIGER-AI-Lab/VLM2Vec#173 · Filter by author Yingshu-Li opened on Nov 11, 2025

• Reproducibility Issue

  TIGER-AI-Lab/VLM2Vec#170 · Filter by author VarunGumma opened on Nov 7, 2025
  1

• Support for Qwen3 and Modeling file Changes

  documentationenhancement
  TIGER-AI-Lab/VLM2Vec#169 · Filter by author VarunGumma opened on Nov 7, 2025
  1

• Train with mmeb in image_only mode, raise error about picke

  Replied
  TIGER-AI-Lab/VLM2Vec#162 · Filter by author saicoco opened on Oct 22, 2025
  1

• More Retriever Implementations for Latest Multi-Vector Models

  TIGER-AI-Lab/VLM2Vec#161 · Filter by author StupidBuluchacha opened on Oct 22, 2025
  2

• Will you update the code repo that supports training with multi-node DeepSpeed?

  Replied
  TIGER-AI-Lab/VLM2Vec#152 · Filter by author xschen-beb opened on Sep 30, 2025
  1

• how to get pos_clip_name

  TIGER-AI-Lab/VLM2Vec#151 · Filter by author LIUYUANWEI98 opened on Sep 23, 2025
  1

• Hard Negative training code release

  Replied
  TIGER-AI-Lab/VLM2Vec#150 · Filter by author FightingFighting opened on Sep 11, 2025
  1

• Significant Performance Gap for the visdoc tasks in VLM2Vec-v2 Qwen2-VL-2B Reproduction

  TIGER-AI-Lab/VLM2Vec#149 · Filter by author FightingFighting opened on Sep 9, 2025
  9

• Why don’t we see the special tokens <|vision_start|> and <|vision_end|> in the input?

  TIGER-AI-Lab/VLM2Vec#147 · Filter by author chataicode opened on Aug 28, 2025
  4

• Discrepancy in Prompt Construction for LamRA Inference

  TIGER-AI-Lab/VLM2Vec#144 · Filter by author L-cuixiao opened on Aug 19, 2025
  1