Yi-Fu Wu comments

Repositories
Issues
Comments

Results 3 comments of


                                            Yi-Fu Wu

chore: Bump vllm to 0.11.2, torch to 2.9, transformers to 4.57.1

> @yfw we need to update the torch version in `tools/build-custom-vllm.sh` as well I pulled in this change for this: https://github.com/NVIDIA-NeMo/RL/pull/1563/commits/eab60198c430cdd2750388951c76e3bc101d64b8

Update megatron bridge and megatron-core

Closes https://github.com/NVIDIA-NeMo/RL/issues/1575

Qwen2.5-vl crash when using mcore backend

Seems this is actually fixed in recent Megatron-Bridge (https://github.com/NVIDIA-NeMo/Megatron-Bridge/commit/d47152e8da4f043f367cf24698373159c017efb0), so we will just need to update the Megatron-Bridge commit we're using