Issues

• 请求更新Qwen3-vl系列的模型

  om-ai-lab/VLM-R1#356 · Filter by author QimingLi-scir opened on Jan 29, 2026
  1

• Fix performance overhead caused by incorrect indentation in image resizing logic

  om-ai-lab/VLM-R1#355 · Filter by author xiaoyang-bi opened on Jan 19, 2026

• multi-image training in GRPO

  om-ai-lab/VLM-R1#354 · Filter by author lapuuu-810 opened on Nov 25, 2025

• ImportError: cannot import name 'Glm4vForConditionalGeneration' from 'transformers'

  om-ai-lab/VLM-R1#353 · Filter by author wja030105-oss opened on Oct 24, 2025
  4

• Agent memory integration

  om-ai-lab/VLM-R1#352 · Filter by author pILLOW-1 opened on Oct 20, 2025

• 加载internvl模型报错

  om-ai-lab/VLM-R1#351 · Filter by author tanjjjjrr opened on Oct 17, 2025

• 请问多久可以支持Qwen3-VL

  om-ai-lab/VLM-R1#350 · Filter by author Henry-Lee-real opened on Oct 15, 2025
  7

• 您好，能够提供一下具体的，可直接运行的ovd任务的训练脚本？感谢。

  om-ai-lab/VLM-R1#348 · Filter by author xiebinghua opened on Oct 5, 2025

• NaN or Inf found in input tensor when using LLaMA-Factory to train the SFT model for REC task.

  om-ai-lab/VLM-R1#346 · Filter by author zaiquanyang opened on Sep 19, 2025

• are you interested using gspo to train a vlm?

  om-ai-lab/VLM-R1#345 · Filter by author wangsujia opened on Sep 19, 2025

• qwenvl没有图文交错了，同时多图输入“论文和宣传的点”实现都有问题啊

  om-ai-lab/VLM-R1#344 · Filter by author NieCha opened on Sep 18, 2025

• Why is the reference model not updated during the entire training cycle?

  om-ai-lab/VLM-R1#343 · Filter by author feivellau opened on Sep 12, 2025

• GUI task multi-image training

  om-ai-lab/VLM-R1#342 · Filter by author songzijian1999 opened on Sep 11, 2025
  4

• About the Coordinates Offsets for Qwen2.5-VL.

  om-ai-lab/VLM-R1#341 · Filter by author Hulmes0217 opened on Sep 6, 2025

• detection_score vs odLength

  om-ai-lab/VLM-R1#339 · Filter by author llf1234 opened on Aug 28, 2025

• 8 3090 GPU LORA OOM issue

  om-ai-lab/VLM-R1#337 · Filter by author Kkkkkz21 opened on Aug 27, 2025
  1

• use_vllm not used in grpo_trainer.py

  om-ai-lab/VLM-R1#336 · Filter by author YongchengYAO opened on Aug 25, 2025

• Grounding DINO Refcoco+ val performance

  om-ai-lab/VLM-R1#335 · Filter by author wawaa opened on Aug 25, 2025

• Will you release object detection with RL?

  om-ai-lab/VLM-R1#334 · Filter by author John1231983 opened on Aug 25, 2025

• GRPO reward collapse?

  om-ai-lab/VLM-R1#332 · Filter by author DidiD1 opened on Aug 8, 2025

• [Bug] Error for Qwen/Qwen2.5-VL-3B-Instruct when using transformers 4.49.0: Input and cos/sin must have the same dtype, got torch.float32 and torch.bfloat16

  om-ai-lab/VLM-R1#331 · Filter by author YongchengYAO opened on Aug 7, 2025

• Will the input data be automatically shuffled?

  om-ai-lab/VLM-R1#330 · Filter by author Shawn-Tao opened on Aug 7, 2025

• [Bug] auto class for ref model: please use model_cls

  om-ai-lab/VLM-R1#328 · Filter by author YongchengYAO opened on Aug 5, 2025

• [Tips] Explicitly define torch_dtype in model loading scripts

  om-ai-lab/VLM-R1#327 · Filter by author YongchengYAO opened on Aug 1, 2025

• [Bug] model_init_kwargs not used in InvernVLModule. get_model_class

  om-ai-lab/VLM-R1#325 · Filter by author YongchengYAO opened on Jul 31, 2025
  1