Search results

[Desync] qwen-omni-utils on GitHub is outdated and out of sync with the PyPI release.

Status: Open.

#391 In QwenLM/Qwen2.5-Omni;

· Tohrusky opened on Mar 14, 2026

KeyError: 'video_fps'

Status: Open.

#390 In QwenLM/Qwen2.5-Omni;

· Antoine-Delplace-Maki opened on Feb 3, 2026

Why is Qwen2.5-Omni an OLM?

Status: Open.

#389 In QwenLM/Qwen2.5-Omni;

· Alexisxty opened on Jan 23, 2026

报错：/pytorch/aten/src/ATen/native/cuda/IndexKernelUtils.cu:16: vectorized_gather_kernel: block: [28,0,0], thread: [127,0,0] Assertion ind >=0 && ind < ind_dim_size && "vectorized gather kernel index out of bounds" failed.

Status: Open.

#387 In QwenLM/Qwen2.5-Omni;

· Bounty-hunter opened on Jan 4, 2026

AWQ 模型 跑不起来

Status: Open.

#386 In QwenLM/Qwen2.5-Omni;

· jujunchen opened on Jan 1, 2026

You are using a model of type qwen2_5_omni_thinker to instantiate a model of type qwen2_5_omni. This is not supported for all configurations of models and can yield errors.

Status: Open.

#385 In QwenLM/Qwen2.5-Omni;

· 2013358072 opened on Dec 23, 2025

Can I use fixed frames during inference?

Status: Open.

#383 In QwenLM/Qwen2.5-Omni;

· wwxs123W opened on Dec 22, 2025

请问如何训练 talker部分？

Status: Open.

#381 In QwenLM/Qwen2.5-Omni;

· JoshonSmith opened on Dec 15, 2025

Reproducing experiment results on MMAU

Status: Open.

#378 In QwenLM/Qwen2.5-Omni;

· qq31415926 opened on Dec 5, 2025

audio video interleave ordering

Status: Open.

#377 In QwenLM/Qwen2.5-Omni;

· wjc2830 opened on Nov 29, 2025

vllm推理think部分报错，assert audio_feature_lengths.shape[0] == 1，不支持batch推理吗

Status: Open.

#376 In QwenLM/Qwen2.5-Omni;

· YuiGao opened on Nov 20, 2025

关于参数的问题

Status: Open.

#375 In QwenLM/Qwen2.5-Omni;

· newforyou opened on Oct 30, 2025