Search results

GRPO vllm_mode=server时，server端报错NotImplementedError

Status: Open.

#6935 In modelscope/ms-swift;

· leuitong opened on Dec 8, 2025

grpo多模态数据训练报错

Status: Open.

#6934 In modelscope/ms-swift;

· TAOSHss opened on Dec 8, 2025

Support sequential data loading (disable shuffle) for Megatron training

Status: Open.

#6932 In modelscope/ms-swift;

· hkxxxxx opened on Dec 7, 2025

Support audio-flamingo-3 or not?

Status: Open.

#6930 In modelscope/ms-swift;

· fengmanlou11 opened on Dec 7, 2025

adapters里填多个adapter会发生什么？

Status: Open.

#6929 In modelscope/ms-swift;

· Leeying9 opened on Dec 7, 2025

qwen3-vl 在全参sft后的模型上做grpo，训练一定step后遇到报错Watchdog caught collective operation timeout

Status: Open.

#6927 In modelscope/ms-swift;

· Dswn888 opened on Dec 7, 2025

使用vllm部署gptqmodel或auto-aptq量化的微调后的qwen3-vl时报错：KeyError: 'layers.0.mlp.down_proj.g_idx'

Status: Open.

#6925 In modelscope/ms-swift;

· Wangbiao2 opened on Dec 6, 2025

qwen3 omni GRPO USE_AUDIO_IN_VIDEO 貌似传不了音频token

Status: Open.

#6924 In modelscope/ms-swift;

· callmeluoq-art opened on Dec 6, 2025

megatron sft下context_parallel_size无效

Status: Open.

#6923 In modelscope/ms-swift;

· daiwc opened on Dec 6, 2025

使用全参sft后的qwen3vl-8b模型继续做grpo时（使用了padding_free），遇到RuntimeError: split_with_sizes expects split_sizes to sum exactly to 1 (input tensor's size at dimension 0), but got split_sizes=[11, 615, 615, 615, 615, 615, 615, 615, 448, 448, 448, 448, 448, 448, 448, 448]

Status: Open.

#6921 In modelscope/ms-swift;

· Dswn888 opened on Dec 6, 2025

Qwen3VL-30A3多图训练参数求助

Status: Open.

#6920 In modelscope/ms-swift;

· jiangsongtao opened on Dec 6, 2025

如何查看eval loss

Status: Open.

#6917 In modelscope/ms-swift;

· microchila opened on Dec 5, 2025