feat: add Qwen3.5 CP support for MCore path#2312

Draft

zpqiu wants to merge 3 commits intoNVIDIA-NeMo:mainfrom

zpqiu:qwen3.5_mcore_cp

Commits on Apr 22, 2026

chore: bump Megatron-Bridge to latest main and Megatron-LM to dev
zpqiu
committed
feat(megatron): delegate packing to the model for VLM + CP (Qwen3.5 VL)
zpqiu
committed
perf: use tuple indexing in CP zigzag slicing to silence PyTorch warning
zpqiu
committed