-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Pull requests: JustVugg/colibri
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
sampling: partial top-p select in dist_build — O(V) heapify + k pops, not O(V log V) qsort (#335)
#354
opened Jul 17, 2026 by
woolcoxm
Contributor
Loading…
docs(env): refresh ENVIRONMENT.md — document all 34 drifted env vars
#353
opened Jul 17, 2026 by
woolcoxm
Contributor
Loading…
fix(test): make test_stops build on Windows (mkdtemp compat shim)
#352
opened Jul 17, 2026 by
woolcoxm
Contributor
Loading…
fix(flake): create flake.lock, use --set-default for COLI_ENGINE
#351
opened Jul 17, 2026 by
LordMZTE
Loading…
4 of 6 tasks
fix(warnings): silence two -Wall/-Wextra warnings on the MinGW build
#350
opened Jul 17, 2026 by
woolcoxm
Contributor
Loading…
fix(omp): park idle workers under PASSIVE policy (was spinning all cores at idle) #341
#349
opened Jul 17, 2026 by
woolcoxm
Contributor
Loading…
tools: rate-scale the E8 lattice ball by bit-width (#81) — int3-e8 is now real int3
#347
opened Jul 17, 2026 by
ZacharyZcR
Contributor
Loading…
cuda: COLI_GROUP_ASYNC=1 — overlap the CPU expert rows with the GPU groups at decode (opt-in, +6-8%)
#342
opened Jul 17, 2026 by
ZacharyZcR
Contributor
Loading…
glm.c: detect 9p via statfs f_type, not the /mnt/ path prefix
#340
opened Jul 17, 2026 by
tonuonu
Loading…
AMD GPU support: HIP/ROCm via single-source backend_gpu_compat.h (+ WMMA dispatch gate)
#339
opened Jul 17, 2026 by
noobdev-ph
Loading…
6 tasks done
GPU backend: failure-path hardening + tests (sticky-error fix, cached-tensor contract, fault-injection hook)
#338
opened Jul 17, 2026 by
noobdev-ph
Loading…
5 of 6 tasks
resource_plan: fix --auto-tier pinning decode to one core (#325)
#332
opened Jul 16, 2026 by
woolcoxm
Contributor
Loading…
st: chunked pread with EINTR retry and honest short-read errors
#331
opened Jul 16, 2026 by
nbeerbower
Loading…
tok: o200k pre-tokenizer support, auto-detected, with download-free test coverage
#330
opened Jul 16, 2026 by
nbeerbower
Loading…
docs: Windows 11 native install walkthrough (toolchain, Smart App Control, CUDA DLL, failure index)
#329
opened Jul 16, 2026 by
mohamedmastouri2000-boop
Loading…
convert_olmoe.py: wire --ebits through to quantization (fixes #323)
#328
opened Jul 16, 2026 by
mohamedmastouri2000-boop
Loading…
Inkling (Thinking Machines 975B MoE): new engine, o200k tokenizer, bf16->int4 converter
#312
opened Jul 16, 2026 by
nbeerbower
Loading…
cuda+engine: full fmt=4 (grouped int4 gs=64) support + diagnostic harness
#298
opened Jul 16, 2026 by
woolcoxm
Contributor
Loading…
4 of 5 tasks
Windows/CUDA port: thermal gate, storage probe, correct dashboard stats
#296
opened Jul 16, 2026 by
David2024patton
Loading…
pipe: blocking pipe_wait (COLI_PIPE_BLOCK=1) + PIPE_WORKERS implies PIPE=1
#270
opened Jul 15, 2026 by
ebootheee
Loading…
feat: implement I/O-aware RAM fallback router and native string-indexed memory
#235
opened Jul 15, 2026 by
dannythevibe
Loading…
Nearly double the hit-rate
#223
opened Jul 14, 2026 by
withinboredom
Contributor
Loading…
4 of 5 tasks
aarch64: add runtime i8mm expert kernel
#217
opened Jul 14, 2026 by
RonitBStudent
Contributor
•
Draft
serve stage 2: response_format → per-request grammars → grammar-forced drafts in the multiplexed server (stacked on #148)
#192
opened Jul 14, 2026 by
fabio-rovai
Contributor
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.