Skip to content

Actions: verl-project/verl

Actions

cpu_unit_tests

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
cpu_unit_tests.yml
6,799 workflow runs
6,799 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

[misc] feat: enable expandable segment support for ascend npu cpu_unit_tests #9330: Pull request #5595 opened by ji-huazhong
20m 32s npu-ci:mem
npu-ci:mem
20m 32s
[trainer] feat: Support CP with TorchtitanEngine cpu_unit_tests #9329: Pull request #5594 synchronize by acisseJZhong
5m 46s acisseJZhong:ep_cp
acisseJZhong:ep_cp
5m 46s
[trainer] feat: Support CP with TorchtitanEngine cpu_unit_tests #9328: Pull request #5594 opened by acisseJZhong
5m 51s acisseJZhong:ep_cp
acisseJZhong:ep_cp
5m 51s
[WIP] feat: Multi-Task On-Policy Distillation cpu_unit_tests #9327: Pull request #5164 synchronize by JacobHelwig
5m 46s JacobHelwig:jhelwig/MOPD
JacobHelwig:jhelwig/MOPD
5m 46s
[vllm,trainer,algo] feat: Enable On-Policy Distillation for VLM cpu_unit_tests #9326: Pull request #5592 synchronize by JacobHelwig
24m 44s JacobHelwig:jhelwig/opdServer
JacobHelwig:jhelwig/opdServer
24m 44s
[fsdp,vllm,trainer,algo] feat: On-Policy Distillation cpu_unit_tests #9325: Pull request #4897 synchronize by JacobHelwig
7m 17s JacobHelwig:jhelwig/onPolicyDistillation
JacobHelwig:jhelwig/onPolicyDistillation
7m 17s
[rollout] feat: configurable token2text field mapping and tokenizer fallback in tracing, cpu_unit_tests #9324: Pull request #5588 synchronize by AnikiFan
Action required AnikiFan:feat/rollout-trace-flexible-token2text
AnikiFan:feat/rollout-trace-flexible-token2text
Action required
[fsdp,vllm,trainer,algo] feat: On-Policy Distillation cpu_unit_tests #9323: Pull request #4897 synchronize by JacobHelwig
16m 8s JacobHelwig:jhelwig/onPolicyDistillation
JacobHelwig:jhelwig/onPolicyDistillation
16m 8s
[vllm,trainer,algo] feat: Enable On-Policy Distillation for VLM cpu_unit_tests #9322: Pull request #5592 synchronize by JacobHelwig
9m 50s JacobHelwig:jhelwig/opdServer
JacobHelwig:jhelwig/opdServer
9m 50s
[fsdp,vllm,trainer,algo] feat: On-Policy Distillation cpu_unit_tests #9321: Pull request #4897 synchronize by JacobHelwig
29m 46s JacobHelwig:jhelwig/onPolicyDistillation
JacobHelwig:jhelwig/onPolicyDistillation
29m 46s
[vllm,trainer,algo] feat: Enable On-Policy Distillation for VLM cpu_unit_tests #9320: Pull request #5592 opened by JacobHelwig
18m 40s JacobHelwig:jhelwig/opdServer
JacobHelwig:jhelwig/opdServer
18m 40s
[megatron] fix: apply override_transformer_config inside mindspeed en… cpu_unit_tests #9319: Commit 592cb07 pushed by ji-huazhong
34m 35s main
main
34m 35s
[fsdp]fix: pass dp_group to prepare_dynamic_batch to fix CUDA deadlock cpu_unit_tests #9318: Pull request #5591 opened by JenniferWang
Action required JenniferWang:fix/fsdp-dynamic-bsz-dp-group
JenniferWang:fix/fsdp-dynamic-bsz-dp-group
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? cpu_unit_tests #9317: 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? cpu_unit_tests #9316: 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? cpu_unit_tests #9315: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[recipe, algo] feat: atropos RL environment integration cpu_unit_tests #9314: Pull request #5514 synchronize by daspartho
Action required daspartho:feat/atropos-integration
daspartho:feat/atropos-integration
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? cpu_unit_tests #9313: 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? cpu_unit_tests #9312: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[vllm] chore: fix mc2 used in vllm_ascend on A2 npu cpu_unit_tests #9311: Pull request #5560 synchronize by wucong25
5m 0s wucong25:wc/fix_a2_mc2
wucong25:wc/fix_a2_mc2
5m 0s
[vllm] chore: fix mc2 used in vllm_ascend on A2 npu cpu_unit_tests #9310: Pull request #5560 synchronize by wucong25
5m 49s wucong25:wc/fix_a2_mc2
wucong25:wc/fix_a2_mc2
5m 49s
[trainer,ckpt,rollout] fix: wake up rollout replicas when actor update is skipped by critic warmup cpu_unit_tests #9309: Pull request #5590 opened by lwzhenglittle
Action required lwzhenglittle:fix-rollout-wake-up-without-update-actor
lwzhenglittle:fix-rollout-wake-up-without-update-actor
Action required
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? cpu_unit_tests #9308: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required
[megatron] fix: apply override_transformer_config inside mindspeed engine to avoid confict with other training engine cpu_unit_tests #9307: Pull request #5589 opened by ChengQianqian
39m 52s ChengQianqian:main
ChengQianqian:main
39m 52s
why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? cpu_unit_tests #9306: Pull request #4263 synchronize by khazic
Action required khazic:main
khazic:main
Action required