Skip to content

Pull requests: JustVugg/colibri

New pull request New
33 Open 322 Closed
33 Open 322 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

docker: slim multi-stage image + compose (alongside the beginner Dockerfile)
#582 opened Jul 24, 2026 by ZacharyZcR Contributor Loading…
serve: single-slot MTP speculation in the mux path (KV_SLOTS=1) — #492/#358
#581 opened Jul 24, 2026 by ZacharyZcR Contributor Loading…
convert: default to group-scaled int4 (gs64) — the #455 tracking item
#579 opened Jul 24, 2026 by ZacharyZcR Contributor Loading…
web: ship the Expert Atlas data (experts.json) — the Brain hover has been fetching a file that never shipped
#578 opened Jul 24, 2026 by ZacharyZcR Contributor Loading…
install: ship version.py + make the launcher survive its absence (#575)
#577 opened Jul 24, 2026 by ZacharyZcR Contributor Loading…
serve: emit finish_reason=length on real cap-hits, not just at admission (addresses a side-bug named in #455)
#576 opened Jul 23, 2026 by NeuralNotwerk Contributor Loading…
feat(olmoe): LFRU prefetch eviction guard to prevent cache thrashing
#571 opened Jul 23, 2026 by EgonRuiter Contributor Loading…
build: portable build_cuda.bat with MSVC/CUDA/arch auto-detection
#570 opened Jul 23, 2026 by ghostnumber7 Loading…
2 of 5 tasks
Fix stateful KV tail at the NGEN limit
#567 opened Jul 23, 2026 by winklemad Loading…
3 of 5 tasks
1
add persistence controls for lower SSD writes
#555 opened Jul 23, 2026 by Skater1808 Loading…
5 tasks
2
CPU: KV cache quantization — KV8 (fp8 e4m3) + KV_TQ (rotated-int4 / PolarQuant)
#553 opened Jul 23, 2026 by NeuralNotwerk Contributor Loading…
3
feat: add WebGPU expert workers
#552 opened Jul 23, 2026 by gauravsaini Loading…
1
feat: add distributed expert workers
#551 opened Jul 23, 2026 by gauravsaini Loading…
1
feat: add dense MLP activation sharding
#550 opened Jul 23, 2026 by gauravsaini Loading…
1
feat: add Qwen3-30B-A3B engine
#544 opened Jul 23, 2026 by opxyc Loading…
4 of 5 tasks
1
fix: convert_olmoe_merged.py no longer needs the whole checkpoint in RAM
#543 opened Jul 23, 2026 by opxyc Loading…
4 of 5 tasks
feat: interactive chat mode for olmoe (multi-turn conversations)
#542 opened Jul 23, 2026 by opxyc Loading…
4 of 5 tasks
perf: add AVX2 int8 dot-product path to olmoe's matmul_q
#541 opened Jul 23, 2026 by opxyc Loading…
4 of 5 tasks
feat: add learned partial mirror planner
#536 opened Jul 22, 2026 by bherald Loading…
1
feat: support OpenAI stop sequences
#535 opened Jul 22, 2026 by bherald Loading…
8
formats: self-describing container stamp (TRUST-VERIFY-REFUSE) + FORMATS.md registry (#524)
#529 opened Jul 22, 2026 by monotophic Contributor Loading…
5
fmt=7: fp8-e4m3-b128 passthrough weight format — CPU + Metal + repack tool (#524)
#528 opened Jul 22, 2026 by monotophic Contributor Loading…
6
docs: soften tiny oracle self-test expectation
#511 opened Jul 22, 2026 by CooperSheroy Loading…
3 of 5 tasks
3
pilot: multi-worker PILOT_REAL prefetch (PILOT_WORKERS) — byte-identical base for the #441 hardware A/B
#480 opened Jul 21, 2026 by cdhdt Contributor • Draft
4 of 5 tasks
4
metal: grouped-int4 (fmt=4) GEMV support + two latent fmt=4 fixes
#457 opened Jul 20, 2026 by monotophic Contributor Loading…
10
ProTip! Updated in the last three days: updated:>2026-07-20.