nemo-mbridge-perf-moe-vlm-training

【效率与协作】Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准