[trainer] feat: add padding for tensor alignment in preprocess_thd_no… #4577
vermouth1992e2e_ppo_trainer_megatron_vllm_2.yml
on: push
setup
9s
e2e_ppo_trainer_megatron-moe-expert-parallel
19m 5s
e2e_ppo_trainer_fsdp_vllm
21m 41s
e2e_ppo_trainer_fsdp-qwen2_5vl-3b
20m 25s
cleanup
5s