-
Notifications
You must be signed in to change notification settings - Fork 190
Pull requests: lightseekorg/tokenspeed
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(scheduler): count active pages across groups
#669
opened Jul 13, 2026 by
lightseek-bot
Contributor
Loading…
fix(prefill-graph): illegal memory access on the flat KV-cache path
#667
opened Jul 13, 2026 by
nperrin-fr
Collaborator
Loading…
feat(kernel): add gfx950 Gluon bf16 MLA decode and prefill kernels
#664
opened Jul 13, 2026 by
Yu-Zhewen
Contributor
Loading…
feat(profile): merge Proton and VizTracer traces onto one timeline
#662
opened Jul 13, 2026 by
antiagainst
Member
•
Draft
perf(deepseek-v4): select two-stage fused mHC launch by token count
#661
opened Jul 13, 2026 by
Xiangyi1996
Collaborator
•
Draft
fix(pd): correct DeepSeek V4 layerwise cache handoff
#660
opened Jul 13, 2026 by
lucifer1004
Contributor
•
Draft
[WIP] test(kernel): verify kernel selection cross-platform and drop expected_kernel_name
#659
opened Jul 13, 2026 by
antiagainst
Member
Loading…
feat(runtime): scheduler-driven full_refresh bit for page-table mirror
#651
opened Jul 12, 2026 by
raikonenfnu
Contributor
Loading…
feat(kernel): add SM120 FlashInfer MXFP4 MoE
#645
opened Jul 11, 2026 by
lucifer1004
Contributor
Loading…
Route MLA RoPE KV writes through embedding API
#643
opened Jul 11, 2026 by
qedawkins
Contributor
Loading…
fix(deepseek-v4): route large-token fused mHC around the allinone cliff
#622
opened Jul 9, 2026 by
Xiangyi1996
Collaborator
•
Draft
perf(deepseek-v4): widen sparse compress cache launch for large ratios
#621
opened Jul 9, 2026 by
Xiangyi1996
Collaborator
Loading…
feat: add DeepSeek V4 L2 KV cache offload and perf optimize.
#620
opened Jul 9, 2026 by
SimonCqk
Contributor
Loading…
fix(spec): restore sentinel padding + context clamps for every drafter
#618
opened Jul 9, 2026 by
jasl
Contributor
Loading…
feat(mla): support custom tree masks in decode
#617
opened Jul 9, 2026 by
mesaleh
Contributor
Loading…
fix(trtllm-mla): make spec-decode CUDA graph capture causal
#616
opened Jul 9, 2026 by
mesaleh
Contributor
Loading…
perf(multimodal): optimize SHM input transport
#610
opened Jul 8, 2026 by
yechank-nvidia
Collaborator
Loading…
perf(multimodal): optimize Qwen vision runtime
#609
opened Jul 8, 2026 by
yechank-nvidia
Collaborator
•
Draft
perf(multimodal): optimize Qwen3.5 M-RoPE preparation
#608
opened Jul 8, 2026 by
yechank-nvidia
Collaborator
•
Draft
refactor(glm): reuse DeepseekV3 base implementation via template methods
#604
opened Jul 7, 2026 by
wtsung
Contributor
Loading…
fix(runtime): restore EnvField overrides after exceptions
#601
opened Jul 7, 2026 by
Xiangyi1996
Collaborator
•
Draft
Previous Next
ProTip!
no:milestone will show everything without a milestone.