Search results

How was patch_embed.proj (Conv3d) initialized from SigLIP2 weights (Conv2d)?

Status: Open.

#2087 In QwenLM/Qwen3-VL;

· ymp5078 opened on Apr 23, 2026

如何选择trainer/How to choose trainer

Status: Open.

#2086 In QwenLM/Qwen3-VL;

· Joeychenj opened on Apr 22, 2026

qwen_vl_utils: _read_video_torchvision misses video_start offset in frames_indices causes incorrect frame index mapping

Status: Open.

#2085 In QwenLM/Qwen3-VL;

· wjm-wjm opened on Apr 17, 2026

fail to reproduce the results on muirbench with qwen3-vl-8b-instruct

Status: Open.

#2083 In QwenLM/Qwen3-VL;

· zhangxy-2019 opened on Apr 15, 2026

5090推理时，报CUDA error: an illegal memory access was encountered

Status: Open.

#2082 In QwenLM/Qwen3-VL;

· cwgxy opened on Apr 7, 2026

lora enable vs set_model in train_qwen.py

Status: Open.

#2079 In QwenLM/Qwen3-VL;

· Arslan-Mehmood1 opened on Mar 28, 2026

使用vllm进行qwen3-vl模型推理时，vllm==0.11.0, qwen_vl_utils=0.0.14,transformers==4.57.0,出现RuntimeError: Numpy is not available，numpy版本是2.2.6

Status: Open.

#2078 In QwenLM/Qwen3-VL;

· Nobelvictory opened on Mar 27, 2026

has anyone fine tuned qwen3-vl for video classification?

Status: Open.

#2077 In QwenLM/Qwen3-VL;

· Arslan-Mehmood1 opened on Mar 26, 2026

Switching attention to Eager causes attention masks to cause IndexError with batch size > 1

Status: Open.

#2076 In QwenLM/Qwen3-VL;

· Apprisco opened on Mar 26, 2026

Input image size limitations

Status: Open.

#2075 In QwenLM/Qwen3-VL;

· UralmashFox opened on Mar 25, 2026

千问3.5-9B模型输出token乱码问题

Status: Open.

#2074 In QwenLM/Qwen3-VL;

· leo-leonardo opened on Mar 25, 2026

多图推理时输出被截断

Status: Open.

#2072 In QwenLM/Qwen3-VL;

· Timeviewerzero opened on Mar 24, 2026