Skip to content

Pull requests: sgl-project/sglang

New pull request New
342 Open 3,756 Closed
342 Open 3,756 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Feature] Support Qserve high priority
#6349 opened May 16, 2025 by HandH1998 Loading…
6 tasks
@zhyncs
1
[fix] illegal memory in _fwd_kernel_ep_scatter_2
#6348 opened May 16, 2025 by xutizhou Loading…
6 tasks
fix: remove content=none test when tool called
#6347 opened May 16, 2025 by shuaills Loading…
6 tasks
Add unique_id_from_bytes for NCCLLibrary
#6343 opened May 16, 2025 by Abatom Loading…
Add fp8 shared_expert kernel for CPU in sgl-kernel and add UT
#6339 opened May 16, 2025 by chunyuan-w • Draft
WIP: [feat] Support different attention backends for prefill and decode
#6338 opened May 16, 2025 by Qiaolin-Yu Loading…
6 tasks
@Fridge003
6
fix: nixl hang
#6335 opened May 15, 2025 by akhoroshev Loading…
1
[Fix] Improve dependencies for Blackwell image
#6334 opened May 15, 2025 by Fridge003 Loading…
6 tasks
@zhyncs
misc: Implement RankZeroFilter for rank-specific logging in model_runner.py
#6333 opened May 15, 2025 by CatherineSue Loading…
3 of 6 tasks
kernel: Remove unnecessary arguments from moe_align_block_size
#6332 opened May 15, 2025 by tanruixiang Loading…
2 of 6 tasks
1
The Gemma template is missing a newline after the user role.
#6331 opened May 15, 2025 by ysulsky Loading…
1 of 6 tasks
fix: allow launch_dummy_health_check_server to start inside of running asyncio loop
#6330 opened May 15, 2025 by ishandhanani Loading…
6 tasks
add fused moe tuning config for Llama-4-Maverick-17B-128E-Instruct wip
#6329 opened May 15, 2025 by BBuf Loading…
6 tasks
1
Refactor DeepSeek logic into atomic operations
#6326 opened May 15, 2025 by fzyzcjy Loading…
6 tasks
4
Refactor DeepSeek MoE layer to unify the two forward branches
#6325 opened May 15, 2025 by fzyzcjy Loading…
6 tasks
12
Minor code cleanup refactor for DeepSeek models
#6324 opened May 15, 2025 by fzyzcjy Loading…
6 tasks
refactor: Extract repeated member variables in KVCache subclasses to base class.
#6323 opened May 15, 2025 by wangxiyu191 Loading…
1 of 6 tasks
Refactor communication logic of DeepSeek for extensibility and understandability
#6321 opened May 15, 2025 by fzyzcjy Loading…
6 tasks
fix ds series model stream bug high priority
#6319 opened May 15, 2025 by sleepcoo Loading…
@CatherineSue
3
fix: limit peak memory usage when computing logprobs
#6318 opened May 15, 2025 by aftersnow Loading…
3 of 6 tasks
Fix one wasted kernel in DeepSeek and minor refactor
#6316 opened May 15, 2025 by fzyzcjy Loading…
6 tasks
[RL] allow weight updation with dp attention enabled
#6311 opened May 15, 2025 by zhuzilin Loading…
1 of 6 tasks
2
[RL] Remove the w13 weight_scale and input_scale for UnquantizedEPMoE…
#6308 opened May 15, 2025 by zhuzilin Loading…
1 of 6 tasks
[Fix] Add undefined update_tensor_inplace function
#6307 opened May 15, 2025 by b8zhong Loading…
[FA][Test] Fix Sparse FA test
#6306 opened May 15, 2025 by b8zhong Loading…
4
ProTip! Follow long discussions with comments:>50.