-
Notifications
You must be signed in to change notification settings - Fork 2.9k
Pull requests: JustVugg/colibri
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(cuda): charge the expert tier real VRAM, not logical bytes (#687)
#1244
opened Aug 26, 2026 by
Unknown-Findout
Contributor
Loading…
perf(sse41): add SSE4.1 fallback tier for olmoe and qwen36 int8 GEMV
#1239
opened Aug 26, 2026 by
jtinbergen
Loading…
feat(experiments): validate reproducible performance records
#1238
opened Aug 26, 2026 by
ZacharyZcR
Contributor
Loading…
fix(qwen36): keep CUDA tier heat adaptive
#1237
opened Aug 26, 2026 by
ZacharyZcR
Contributor
Loading…
feat(planner): emit evidence-backed next actions
#1236
opened Aug 26, 2026 by
ZacharyZcR
Contributor
Loading…
fix(telemetry): report Vulkan expert residency
#1235
opened Aug 26, 2026 by
ZacharyZcR
Contributor
Loading…
fix(cuda): pin speculative expert kernel family
#1234
opened Aug 26, 2026 by
ZacharyZcR
Contributor
Loading…
fix(qwen36): widen jb buffers, add size_t casts, fix nested comment
#1232
opened Aug 25, 2026 by
Petsku01
Loading…
build(cuda): NVCC_ALLOW_UNSUPPORTED, so a host compiler newer than VS2022 is reachable
#1231
opened Aug 25, 2026 by
Unknown-Findout
Contributor
Loading…
fix(k3): K3_VALIDATE_OUT defaults to a hardcoded /tmp path
#1230
opened Aug 25, 2026 by
Unknown-Findout
Contributor
Loading…
fix(tools): k3_tokenizer --ctest writes its fixture to a hardcoded /tmp path
#1229
opened Aug 25, 2026 by
Unknown-Findout
Contributor
Loading…
feat(glm52): optimize inference on single RTX 3090 (24GB) via Direct I/O and MTP
#1177
opened Aug 22, 2026 by
up1t3
Loading…
feat: fmt=8 (fp8-e4m3) decode on the kv_b absorb path, CPU and CUDA
#1102
opened Aug 19, 2026 by
monotophic
Contributor
Loading…
fix(doctor,planner): make DeepSeek-V4 checkpoints pass doctor and plan correctly
#1006
opened Aug 14, 2026 by
RobinHoodO
Loading…
cuda: zero-copy expert views on pageable-shared memory (GB10)
#936
opened Aug 11, 2026 by
Nanetnounou
Contributor
Loading…
cuda: warp-per-row E8 kernels, lattice table in shared memory
#935
opened Aug 11, 2026 by
Nanetnounou
Contributor
Loading…
Scan vulkan devices in reverse order, to avoid main GPU
discussion
Proposta / discussione aperta, non un task
vulkan
Backend Vulkan/AMD
#917
opened Aug 10, 2026 by
xxxajk
Loading…
3 of 5 tasks
feat(moe): add DEGRADE_ZERO miss-slot zero-fill policy (issue #865)
enhancement
New feature or request
performance
Velocità / tok-s / ottimizzazioni
quality
Qualità del modello / quantizzazione
#906
opened Aug 9, 2026 by
kritikagarg
Loading…
2 of 5 tasks
fix: report Vulkan expert residency in telemetry and brain map
bug
Difetto verificato nel codice
needs-rebase
Confligge, serve rebase dell'autore
vulkan
Backend Vulkan/AMD
#891
opened Aug 8, 2026 by
MasterCATZ
Loading…
Feat/dstorage transport
enhancement
New feature or request
performance
Velocità / tok-s / ottimizzazioni
#825
opened Aug 4, 2026 by
khalilswdp
Contributor
Loading…
4 of 5 tasks
managed runner, deterministic benchmark, and causal evidence (PR3 of #377 split)
discussion
Proposta / discussione aperta, non un task
enhancement
New feature or request
performance
Velocità / tok-s / ottimizzazioni
#822
opened Aug 4, 2026 by
BColsey
Loading…
Previous Next
ProTip!
What’s not been updated in a month: updated:<2026-07-27.