forked from sgl-project/sglang
-
Notifications
You must be signed in to change notification settings - Fork 0
Pull requests: vincentzed/sglang
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[Feature] Decode context parallelism (DCP) for DSA models (DeepSeek V3.2, GLM-5.x)
#13
opened Jul 18, 2026 by
vincentzed
Owner
Loading…
[DSA] Port Inkling symm-mem custom AR family: env-gated drop-in + bit-identical fused decode AR+RMSNorm
#12
opened Jul 16, 2026 by
vincentzed
Owner
Loading…
[M4] Step-1 NO-GO: green-context SM overlap of DSA indexer ∥ main-attention projection loses to serial
#11
opened Jul 8, 2026 by
vincentzed
Owner
Loading…
[DSA] Opt-in MXFP4 index-K cache for GLM-5.x / DeepSeek-V3.2 — fused-kernel MXFP4 mode (port+extension of #26209)
#10
opened Jul 7, 2026 by
vincentzed
Owner
Loading…
docs(agent-tasks): DSA perf workstream index — N4/N7 + M3–M7/M10 hand-off plans
#9
opened Jul 7, 2026 by
vincentzed
Owner
Loading…
JetSpec tree drafting on DFlash — lossless; width=1 clean drop-in
#8
opened Jun 30, 2026 by
vincentzed
Owner
Loading…
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.