[4/n][trainer] feat: flowgrpo - add diffusers + fsdp engine support (#5802)

[trainer] fix: handle empty response_mask in calculate_debug_metrics (#5860)

[megatron] fix: support critic model (#5870)

[ci] fix: rename fsdp-vlm to megatron-vlm in trtllm cleanup needs (#5880)

[tool] feat: verl integrate msprobe data collection (#5186)

[megatron, cfg] feat: add Qwen3.5-122B Megatron launch script (#5874)

[doc] feat: add NVFP4 QAT documentation (#5861)

[docker, ci] fix: all CIs, transformers upgrade to 5.3.0 and vllm==0.18.0 (#5724)

[doc] chore: add npu faq doc (#5871)

[fully_async] chore: Update fully async dapo qwen3-30b npu script (#5864)

[cfg] refactor: unify ppo_trainer and ppo_megatron_trainer config (#5848)

[vllm] fix: Fix vLLM synchronization error caused by SGLang skipping resume optimize (#5866)

[misc] fix: license for verl/workers/rollout/trtllm_rollout/__init__.py (#5862)

[rollout, trtllm] fix: add missing __init__.py to trtllm_rollout package (#5857)

[misc] feat: add agent instructions, skills & improve CI for easier tests (#5846)