Skip to content

Actions: verl-project/verl

Actions

e2e_transferqueue

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
e2e_transferqueue.yml
2,392 workflow runs
2,392 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

[rollout, data] fix: honor train_max_samples/val_max_samples in fully… e2e_transferqueue #3184: Commit c3e3970 pushed by ArronHZG
1h 21m 57s main
main
1h 21m 57s
[tool] fix: handle empty image inputs in ToolAgentLoop (#5420) e2e_transferqueue #3183: Commit 9dd447e pushed by wuxibin89
1h 5m 11s main
main
1h 5m 11s
[trainer] feat: add support for the GDPO algorithm e2e_transferqueue #3182: Pull request #5409 synchronize by yue-zeng-yue
Action required yue-zeng-yue:feat-gdpo
yue-zeng-yue:feat-gdpo
Action required
[BREAKING][rollout,cfg] refactor: get rid of actor_rollout_ref config from rollout e2e_transferqueue #3181: Pull request #5418 synchronize by wuxibin89
1h 44m 57s wuxibin89:wuxibin/refactor_config
wuxibin89:wuxibin/refactor_config
1h 44m 57s
[misc] feat: Add code for data grouping in no-padding scenario e2e_transferqueue #3180: Pull request #5424 synchronize by Kite0011
58m 44s Kite0011:add_rm_support
Kite0011:add_rm_support
58m 44s
[rollout] feat: Fix partial load problem, Add vlm support for trtllm rollout e2e_transferqueue #3177: Pull request #5149 synchronize by SchumiDing
Action required SchumiDing:vlm_trtllm_support
SchumiDing:vlm_trtllm_support
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue #3176: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[rollout] feat: Fix partial load problem, Add vlm support for trtllm rollout e2e_transferqueue #3175: Pull request #5149 synchronize by SchumiDing
Action required SchumiDing:vlm_trtllm_support
SchumiDing:vlm_trtllm_support
Action required
[fsdp] fix: make GDN work correctly when using ulysses and varlen sequence e2e_transferqueue #3174: Pull request #5346 synchronize by nono-Sang
Action required nono-Sang:gdn_fix
nono-Sang:gdn_fix
Action required
[BREAKING][megatron] feat: support Megatron-FSDP as a new training backend e2e_transferqueue #3173: Pull request #5423 opened by conver334
Action required conver334:megatron-fsdp
conver334:megatron-fsdp
Action required
[WIP][algo] Migrate and implement the GDPO algorithm into the existing framework. e2e_transferqueue #3172: Pull request #5422 synchronize by Rhetee
Action required Rhetee:GDPO
Rhetee:GDPO
Action required
[megatron, vllm] feat: NVFP4 (W4A16) QAT training support via ModelOpt e2e_transferqueue #3171: Pull request #5254 synchronize by jQizhang
Action required jQizhang:nvfp4_cleanup
jQizhang:nvfp4_cleanup
Action required
[WIP][algo] Migrate and implement the GDPO algorithm into the existing framework. e2e_transferqueue #3170: Pull request #5422 opened by Rhetee
Action required Rhetee:GDPO
Rhetee:GDPO
Action required
[tool] feature: scheduling analysis based on profiling data for torch profiler e2e_transferqueue #3169: Pull request #5367 synchronize by Rhetee
Action required Rhetee:main
Rhetee:main
Action required
[rollout] feat: Fix partial load problem, Add vlm support for trtllm rollout e2e_transferqueue #3168: Pull request #5149 synchronize by SchumiDing
Action required SchumiDing:vlm_trtllm_support
SchumiDing:vlm_trtllm_support
Action required
[megatron, vllm] feat: NVFP4 (W4A16) QAT training support via ModelOpt e2e_transferqueue #3167: Pull request #5254 synchronize by jQizhang
Action required jQizhang:nvfp4_cleanup
jQizhang:nvfp4_cleanup
Action required
[perf, trtllm] feat: Add Nsight support for rollout server mode (trtllm) e2e_transferqueue #3166: Pull request #5391 synchronize by davidmlw
1h 11m 29s joyang-nv:liweim/nsys
joyang-nv:liweim/nsys
1h 11m 29s
[perf, trtllm] feat: Add Nsight support for rollout server mode (trtllm) e2e_transferqueue #3165: Pull request #5391 synchronize by davidmlw
4m 5s joyang-nv:liweim/nsys
joyang-nv:liweim/nsys
4m 5s
[rollout] refactor: bucketed transfer utils e2e_transferqueue #3164: Pull request #5309 synchronize by pengwu22
1h 45m 28s pengwu22:pw/test-wt-utils
pengwu22:pw/test-wt-utils
1h 45m 28s
[trainer] feat: add padding for tensor alignment in preprocess_thd_no… e2e_transferqueue #3163: Commit 32705dc pushed by vermouth1992
1h 46m 28s main
main
1h 46m 28s
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? e2e_transferqueue #3162: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[megatron, vllm] feat: NVFP4 (W4A16) QAT training support via ModelOpt e2e_transferqueue #3161: Pull request #5254 synchronize by jQizhang
Action required jQizhang:nvfp4_cleanup
jQizhang:nvfp4_cleanup
Action required
[perf] fix: the overwritten of Torch_profile with multi steps. (#5395) e2e_transferqueue #3160: Commit 5f7c345 pushed by wuxibin89
34m 15s main
main
34m 15s
[rollout] refactor: bucketed transfer utils e2e_transferqueue #3159: Pull request #5309 synchronize by pengwu22
35m 34s pengwu22:pw/test-wt-utils
pengwu22:pw/test-wt-utils
35m 34s
[Megatron] feat: Support routing replay on NPU with performance and compatibility enhancements e2e_transferqueue #3158: Pull request #5298 synchronize by 755651978
1h 11m 20s 755651978:main-0212
755651978:main-0212
1h 11m 20s