-
Notifications
You must be signed in to change notification settings - Fork 317
Pull requests: radixark/miles
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Rewrite fully-async rollout as FullyAsyncRolloutFn on the class-based rollout API
#1717
opened Jul 18, 2026 by
yueming-yuan
Collaborator
Loading…
Move fully-async rollout from examples into miles/rollout
#1716
opened Jul 18, 2026 by
yueming-yuan
Collaborator
Loading…
fix(rollout): recover session rollback on router-timeout desync
#1714
opened Jul 18, 2026 by
layahaasini
Loading…
fix(tito): tokenize appended messages as one suffix
run-ci-sglang
#1712
opened Jul 17, 2026 by
guapisolo
Collaborator
Loading…
openenv/tbench2: orphan TTL + ownership labels for per-task sandboxes
#1711
opened Jul 17, 2026 by
nblintao
Loading…
openenv/tbench2: per-task sandbox image recipe + Daytona materialization
#1710
opened Jul 17, 2026 by
nblintao
Loading…
openenv_swebench: SWE-bench-style agentic RL adapter (mirrors TB2)
#1705
opened Jul 17, 2026 by
Shi-Dong
Contributor
Loading…
3 tasks
rollout: support consistent_hashing routing and manual policy with min_load in the inference_rollout stack
#1703
opened Jul 17, 2026 by
yueming-yuan
Collaborator
Loading…
[DO NOT MERGE] ci: negative validation of the metric-history gate
bypass-fastfail
Disable CI fast-fail: run all stages and all tests to surface every failure
run-ci-replay
Routing / indexer replay tests
run-ci-weight-update
Run weight update tests
docs: correct weight-sync mode and other inaccuracies in Quick Start
#1698
opened Jul 16, 2026 by
Shi-Dong
Contributor
Loading…
1 task
docs(async): document dynamic global batch pinning num_steps_per_rollout=1
#1696
opened Jul 16, 2026 by
Shi-Dong
Contributor
Loading…
1 task done
Add GLM-4.7-Flash GSM8K reasoning RL launcher + example doc
#1694
opened Jul 16, 2026 by
Shi-Dong
Contributor
Loading…
2 tasks done
[router] set manual policy (sticky + min_load) as default agentic routing policy
#1690
opened Jul 15, 2026 by
yueming-yuan
Collaborator
Loading…
fix(weight-update): propagate background P2P failures
#1687
opened Jul 15, 2026 by
Xinyu-Kang
Contributor
Loading…
feat(megatron): add after-train-step hook
#1682
opened Jul 15, 2026 by
waitong94
Contributor
Loading…
Add --async-max-concurrent-samples to decouple fully-async generation concurrency from batch size
#1677
opened Jul 15, 2026 by
yueming-yuan
Collaborator
Loading…
openenv/tbench2: per-task Daytona cloud-sandbox execution backend
#1675
opened Jul 14, 2026 by
nblintao
Loading…
fix(args): skip Megatron validate_args for external SGLang server
#1674
opened Jul 14, 2026 by
surega21
Loading…
Backfill async rollouts on sample completions
#1673
opened Jul 14, 2026 by
ys-2020
Contributor
Loading…
openenv/tbench2: DeepSeek-V4-Flash launcher variant
#1662
opened Jul 14, 2026 by
Shi-Dong
Contributor
Loading…
1 task
fix(ci): enable ft-long fault-tolerance soak tests
run-ci-ft-long
#1653
opened Jul 13, 2026 by
guapisolo
Collaborator
Loading…
Previous Next
ProTip!
Add no:assignee to see everything that’s not assigned.