-
Notifications
You must be signed in to change notification settings - Fork 4
Pull requests: QuixiAI/SlimServe
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
GLM-5.3-Flash Q2 (antirez GGUF) served on Metal: KDA, sparse NoPE-MLA, pooled indexer, nextn MTP drafter, texture-unit iq2_xxs codebook - 1.5x ds4 decode, 1.8x ds4 prefill on the same Mac
#30
opened Sep 15, 2026 by
auroter
Contributor
Loading…
GLM-5.3-Flash NVFP4 on 4x RTX PRO 6000 (sm_120): rtx6000 platform, glm53f-nvfp4-4 record, decode and prefill kernels
#29
opened Sep 13, 2026 by
auroter
Contributor
Loading…
metal: harden kernels against the QuixiCore-Metal PR #3 review findings
#23
opened Sep 2, 2026 by
auroter
Contributor
Loading…
metal: vendored README states the real sync contract
#22
opened Sep 1, 2026 by
auroter
Contributor
Loading…
XPU: serve Qwen3.8-27B-NVFP4 on 4x Arc Pro B70 at TP4, at parity with the reference fork
#21
opened Sep 1, 2026 by
ropoctl
Contributor
Loading…
ProTip!
Exclude everything labeled
bug with -label:bug.