-
Notifications
You must be signed in to change notification settings - Fork 72
Pull requests: warpfront/hipfire
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
ENH(qwen35): MTP speculative decode on dense TP through the shared draft and accept path
#769
opened Sep 16, 2026 by
alpineQ
Contributor
Loading…
13 of 17 tasks
MQ4-Lloyd (qt=52): per-tensor Lloyd codebook in the MQ4V2 container, -15% KLD at 4 bpw; mq4-pro prefill fix; AWQ rmsnorm rewrite
ratchet-raise
Maintainer approval for a declared RATCHET-RAISE in this PR
#768
opened Sep 15, 2026 by
Kaden-Schutt
Collaborator
Loading…
fix(mtp): chunk scalar Q8 batched GEMM above 64 rows
#767
opened Sep 15, 2026 by
HUSRCF
Contributor
Loading…
10 of 18 tasks
perf(gfx1201): split packed-Q8 verifier KV scans — +22.6% DFlash E2E at 21.5K context
#760
opened Sep 14, 2026 by
HUSRCF
Contributor
Loading…
perf(dflash): enable validated S1-S6 launch fusions on gfx1201
#759
opened Sep 14, 2026 by
HUSRCF
Contributor
Loading…
6 of 14 tasks
refactor: execute sealed MoE through shared steps
ratchet-raise
Maintainer approval for a declared RATCHET-RAISE in this PR
#755
opened Sep 12, 2026 by
fivetide
Collaborator
Loading…
Escha-W2 port: 2-bit trellis decoded in the GEMV — Qwen3.6-35B-A3B and Qwen3.8-27B
meta:wip
Open, but WIP PRs
#694
opened Sep 4, 2026 by
nwoolmer
Collaborator
Loading…
ProTip!
Follow long discussions with comments:>50.