Skip to content

Actions: verl-project/verl

Actions

e2e_transferqueue_ascend

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
e2e_transferqueue_ascend.yml
606 workflow runs
606 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

[recipe] refactor: refactor ray trainer for separate recipe use. (fully async / one step off) e2e_transferqueue_ascend #37: Pull request #5184 synchronize by ArronHZG
1h 0m 46s meituan-search:refactor_fully_async_ray_trainer
meituan-search:refactor_fully_async_ray_trainer
1h 0m 46s
[algo] feat: support rollout router replay in MegatronEngine e2e_transferqueue_ascend #36: Pull request #5185 synchronize by xhx1022
8m 51s xhx1022:xhx/r3_engine
xhx1022:xhx/r3_engine
8m 51s
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue_ascend #35: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue_ascend #34: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[algo] feat: support rollout router replay in MegatronEngine e2e_transferqueue_ascend #33: Pull request #5185 synchronize by xhx1022
28m 57s xhx1022:xhx/r3_engine
xhx1022:xhx/r3_engine
28m 57s
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue_ascend #32: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[megatron] fix: add protections for logits_processor_args.pop("loss_m… e2e_transferqueue_ascend #31: Commit ad404cb pushed by ISEEKYAN
1h 0m 45s main
main
1h 0m 45s
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue_ascend #30: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue_ascend #28: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[worker, training_utils] fix: Metric Aggregation Across DP Ranks (#5203) e2e_transferqueue_ascend #27: Commit 2320603 pushed by wuxibin89
59m 58s main
main
59m 58s
[BREAKING][reward] refactor: remove reward model worker code and invocation e2e_transferqueue_ascend #26: Pull request #5194 synchronize by yyDing1
32m 6s yyDing1:rm_rm_config
yyDing1:rm_rm_config
32m 6s
[worker, training_utils] fix: Metric Aggregation Across DP Ranks e2e_transferqueue_ascend #25: Pull request #5203 synchronize by JacobHelwig
21m 0s JacobHelwig:jhelwig/fixDPMetrics
JacobHelwig:jhelwig/fixDPMetrics
21m 0s
[worker, training_utils] fix: Metric Aggregation Across DP Ranks e2e_transferqueue_ascend #24: Pull request #5203 synchronize by JacobHelwig
3m 26s JacobHelwig:jhelwig/fixDPMetrics
JacobHelwig:jhelwig/fixDPMetrics
3m 26s
[worker, training_utils] fix: Metric Aggregation Across DP Ranks e2e_transferqueue_ascend #23: Pull request #5203 synchronize by JacobHelwig
4m 53s JacobHelwig:jhelwig/fixDPMetrics
JacobHelwig:jhelwig/fixDPMetrics
4m 53s
[worker, training_utils] fix: Metric Aggregation Across DP Ranks e2e_transferqueue_ascend #22: Pull request #5203 opened by JacobHelwig
4m 17s JacobHelwig:jhelwig/fixDPMetrics
JacobHelwig:jhelwig/fixDPMetrics
4m 17s
[vllm] fix: ignore MoE router layers for FP8 quantization (#5107) e2e_transferqueue_ascend #21: Commit 597b63f pushed by ISEEKYAN
20m 43s main
main
20m 43s
[recipe] refactor: refactor ray trainer for separate recipe use. (fully async / one step off) e2e_transferqueue_ascend #20: Pull request #5184 synchronize by ArronHZG
39m 38s meituan-search:refactor_fully_async_ray_trainer
meituan-search:refactor_fully_async_ray_trainer
39m 38s
[recipe] refactor: refactor ray trainer for separate recipe use. (fully async / one step off) e2e_transferqueue_ascend #19: Pull request #5184 synchronize by ArronHZG
12m 51s meituan-search:refactor_fully_async_ray_trainer
meituan-search:refactor_fully_async_ray_trainer
12m 51s
[ckpt] feat: add mooncake backend e2e_transferqueue_ascend #18: Pull request #5176 synchronize by x1314aq
Action required x1314aq:mooncake
x1314aq:mooncake
Action required
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #17: Pull request #4978 synchronize by daikang6
34m 51s daikang6:main
daikang6:main
34m 51s
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #15: Pull request #4978 synchronize by daikang6
20m 47s daikang6:main
daikang6:main
20m 47s
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #13: Pull request #4978 synchronize by daikang6
21m 33s daikang6:main
daikang6:main
21m 33s
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #12: Pull request #4978 synchronize by daikang6
39m 44s daikang6:main
daikang6:main
39m 44s
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #11: Pull request #4978 synchronize by daikang6
53m 43s daikang6:main
daikang6:main
53m 43s
[ci] feat: add three npu workflow yml test e2e_transferqueue_ascend #10: Pull request #4978 synchronize by daikang6
53m 38s daikang6:main
daikang6:main
53m 38s
ProTip! You can narrow down the results and go further in time using created:<2026-01-29 or the other filters available.