Search results

训练随机中断

Status: Open.

#6268 In modelscope/ms-swift;

· lussssep opened on Oct 23, 2025

用ms-swift训练完多模态模型，之后用vllm（非ms-swift内的vllm）运行此模型，发现指标差异较大，是推理模板不匹配的问题吗？

Status: Open.

#6267 In modelscope/ms-swift;

· pikepokenew opened on Oct 23, 2025

Qwen3 VL 8B Instruct多图sft异常

Status: Open.

#6266 In modelscope/ms-swift;

· zwOnline0519 opened on Oct 23, 2025

自定义模型最佳实践

Status: Open.

#6265 In modelscope/ms-swift;

· aJupyter opened on Oct 22, 2025

Best Practices for RL Training of Qwen3-VL-30B-A3B and Qwen3-VL-235B

Status: Open.

#6264 In modelscope/ms-swift;

· jhl-Det opened on Oct 22, 2025

[Megatron-Swift] <code>--async_save</code> and <code>--use_persistent_ckpt_worker</code> cause <code>remaining_argv</code> error in SFT training

Status: Open.

#6262 In modelscope/ms-swift;

· happy-fishingman opened on Oct 22, 2025

swift怎么把COT的数据根据qwen3的形式进行训练

Status: Open.

#6261 In modelscope/ms-swift;

· xucx666 opened on Oct 22, 2025

Deepseek-OCR多机训练报错

bugSomething isn't workingSomething isn't working

Status: Open.

#6260 In modelscope/ms-swift;

· perdone opened on Oct 22, 2025

部署Qwen2.5-14B-Instruct，在工具调用场景，推理的时候经常会出现乱码&lt;|im_start|&gt;&lt;|im_start|&gt;&lt;|im_start|&gt;&lt;|im_start|

Status: Open.

#6258 In modelscope/ms-swift;

· duoyw opened on Oct 22, 2025

Do we support Agentic RL?

Status: Open.

#6256 In modelscope/ms-swift;

· huaiyizhao opened on Oct 22, 2025

Any plans to support fine-tuning omnivinci?

Status: Open.

#6255 In modelscope/ms-swift;

· King-king424 opened on Oct 22, 2025

我用dify配置本地的rerank模型一直报这个错

Status: Open.

#6248 In modelscope/ms-swift;

· nizuzihengui99 opened on Oct 21, 2025