-
Notifications
You must be signed in to change notification settings - Fork 1.2k
Pull requests: THUDM/slime
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(agent): keep agent turns trainable under reasoning_content echo divergence
#2335
opened Aug 27, 2026 by
JasonFooAGI
Loading…
fix(observability): support W&B 0.28.2 run IDs
#2333
opened Aug 26, 2026 by
cyyecao-lappland
Loading…
Order pipeline weight-update process groups across rollout engines
#2331
opened Aug 26, 2026 by
chengcuiping
Loading…
perf(agent): isolate prompt rendering from event loop
#2328
opened Aug 26, 2026 by
liuhao-labs
Contributor
Loading…
fix(glm5): guard TileLang sparse MLA merge pass
#2313
opened Aug 24, 2026 by
liuhao-labs
Contributor
Loading…
fix(args): reject unsupported colocated offload sync
#2311
opened Aug 24, 2026 by
liuhao-labs
Contributor
Loading…
fix(megatron): compact actor logits before projection
#2310
opened Aug 24, 2026 by
liuhao-labs
Contributor
Loading…
perf(rollout): reuse filtered prompt token ids
#2309
opened Aug 23, 2026 by
liuhao-labs
Contributor
Loading…
fix: distribute routing replay padding across experts
#2308
opened Aug 22, 2026 by
EazyReal
Contributor
Loading…
fix(update-weight): gather legacy TE grouped experts
#2307
opened Aug 22, 2026 by
primorLee
Loading…
feat(rollout): 为异步 RS 精确补齐 prompt-group batch
#2305
opened Aug 21, 2026 by
inmnin
Loading…
10 tasks done
feat(megatron): report peak memory for log_probs and actor_train
#2304
opened Aug 21, 2026 by
EazyReal
Contributor
Loading…
docs: add SkyPilot launch tutorial under platform_support
#2303
opened Aug 21, 2026 by
ishankaul1
Loading…
Previous Next
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.