Morph MORPH
SIGN IN SIGN UP
dhruvhead / miles UNCLAIMED
0 0 0 Python

ACTIVITY

Commits Reviews
week month year
COMMITS
20
in the last week
CONTRIBUTORS
8
active
STARS
0
total
FORKS
0
total
TOP CONTRIBUTORS
Y
Yueming Yuan
9 commits
M
maocheng23
4 commits
Z
Zhichen Zeng
2 commits
J
Jiajun Li
1 commit
S
Shi-Dong
1 commit
H
Huapeng Zhou
1 commit
D
Douglas Yang
1 commit
J
JensenFire
1 commit
RECENT COMMITS
Z
feat: real cp support with relayout fix for qwen3.5 train/rollout mismatch (#885)
Zhichen Zeng
eb294e3
Y
[fix] tiny fix debug rollout only in weight version check (#967)
Yueming Yuan
02f6e05
M
Add skiplist and more robust calculation on val (#965)
maocheng23
8d66ac1
Y
[v0.5.10][7][FSDP] move FSDP to experimental and disable by default (#961)
Yueming Yuan
c4e50c8
Y
[v0.5.10][6][FSDP] fix outdated weight update logic in FSDP (#948)
Yueming Yuan
d6158f8
Y
[v0.5.10][5] Disable piecewise cuda graph to avoid NVLS oom (#935)
Yueming Yuan
c74392d
Y
[v0.5.10][4] Fix _no_split_modules set not subscriptable in transformers >=5.0 (#931)
Yueming Yuan
2a99108
Y
[v0.5.10][3] Fix processor return_tensors duplicate kwarg for transformers >=5.0 (#927)
Yueming Yuan
b1a4346
Y
[v0.5.10][2] Fix apply_chat_template behavior for transformers >=5.0 (#926)
Yueming Yuan
ef228e6
Y
[v0.5.10][1] Bump sglang to v0.5.10 (#898)
Yueming Yuan
4d8b007
M
Fix/pause generation mode (#924)
maocheng23
94dbb8f
M
Add weight staleness control for fully async rollout (#958)
maocheng23
41615af
J
chore: remove obsolete swe-agent server.py and run-qwen3.sh (#952)
Jiajun Li
6b58ebd
S
Expose rollout-batch-size, n-samples-per-prompt, global-batch-size as CLI args in swe-agent-v2 (#954)
Shi-Dong
4db9bfe
H
[example] add retool v2 example with multi-turn framework interfaces (#654)
Huapeng Zhou
afc5b55
Y
Add consistent hashing routing policy for rollout (#891)
Yueming Yuan
c198efa
D
fix: adding thin glm5 image to docker build + latest tag sync (#871)
Douglas Yang
70dc402
M
Add heartbeat and id to session server (#866)
maocheng23
eaa36a2
J
[BUGFIX] Invoke "post_process_quantization" by default after weight updating (#890)
JensenFire
8146a78
Z
Bump flash-linear-attention version to 0.4.2 (#892)
Zhichen Zeng
3572922