-
Notifications
You must be signed in to change notification settings - Fork 3.4k
Pull requests: sgl-project/sglang
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix accept rate in speculative decoding metrics
#13212
opened Nov 13, 2025 by
SiqiLi-Fighting
Loading…
docs: update fused MoE config path
documentation
Improvements or additions to documentation
#13211
opened Nov 13, 2025 by
edwardzjl
Loading…
4 tasks
Fix broken Markdown formatting in DeepEP documentation
documentation
Improvements or additions to documentation
#13210
opened Nov 13, 2025 by
Taishi-N324
Loading…
4 tasks
router select dp group with the minimum number of tokens
router
#13208
opened Nov 13, 2025 by
jiashaokun-1
Loading…
4 tasks
[Bugfix] Add weak_ref_tensor.cpp to package-data
dependencies
Pull requests that update a dependency file
#13204
opened Nov 13, 2025 by
yudian0504
Loading…
4 tasks
Use dynamically maintained num_waiting_tokens in get_load()
#13203
opened Nov 13, 2025 by
vipwangerxiao
Loading…
1 of 4 tasks
Fix bug: Incorrect variable used in rem_total_token_offset calculatio…
#13201
opened Nov 13, 2025 by
liuhuijiayou
Loading…
[Performance] Move the contiguous to torch compile region
#13199
opened Nov 13, 2025 by
DarkSharpness
Loading…
4 tasks
fix bench_speculative bug
speculative-decoding
#13197
opened Nov 13, 2025 by
Lzhang-hub
Loading…
4 tasks
Feature/patch v0.5.3.post2 giga
amd
dependencies
Pull requests that update a dependency file
documentation
Improvements or additions to documentation
#13196
opened Nov 13, 2025 by
kirilica-zxr
Loading…
4 tasks
Add turbomind_rms_norm to accelerate QK norm in Qwen3 models
high priority
run-ci
sgl-kernel
#13189
opened Nov 13, 2025 by
cscyuge
Loading…
3 of 4 tasks
Update GDN causal conv1d cuda kernel - prepare for new changes
run-ci
sgl-kernel
#13188
opened Nov 13, 2025 by
byjiang1996
Loading…
4 tasks done
Remove unnecessary code which might occupy 500mb memory
run-ci
#13185
opened Nov 13, 2025 by
hebiao064
Loading…
4 tasks
Fix cases where modelscope id and huggginface id don't match and 'SGLANG_USE_MODELSCOPE' is on.
#13184
opened Nov 13, 2025 by
Oklahomawhore
Loading…
4 tasks done
Add FP32 dtype support for RoPE
run-ci
sgl-kernel
#13181
opened Nov 13, 2025 by
jinyouzhi
Loading…
2 of 4 tasks
diffusion: support sp for image models
run-ci
#13180
opened Nov 13, 2025 by
mickqian
Loading…
4 tasks
[Piecewise CUDA Graph] Support W4A8
piecewise-cuda-graph
run-ci
#13179
opened Nov 13, 2025 by
b8zhong
Loading…
fix(multimodal_gen): Fix Wan2.1-T2V-14B model loading and inference
#13174
opened Nov 13, 2025 by
jy-song-hub
Loading…
4 tasks done
Previous Next
ProTip!
no:milestone will show everything without a milestone.