Skip to content

Pull requests: Luce-Org/lucebox

New pull request New
59 Open 444 Closed
59 Open 444 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

feat(moe): add a generic heterogeneous stage planner
#577 opened Aug 3, 2026 by davide221 Contributor • Draft
perf(ds4): overlap q5 verification and stabilize 16K
#575 opened Aug 2, 2026 by davide221 Contributor • Draft
feat(kimi-k3): add DSpark Q8 artifact and shared speculative runtime
#573 opened Aug 1, 2026 by davide221 Contributor • Draft
2
feat(vision): native mmproj multimodal chat for Qwen35
#571 opened Aug 1, 2026 by davidmroth Contributor Loading…
6 tasks
54
feat(ds4): run CUDA and HIP expert parallelism in one process
#570 opened Aug 1, 2026 by davide221 Contributor Loading…
8
perf(ds4): optimize q5 fused verifier on AMD
#569 opened Aug 1, 2026 by davide221 Contributor • Draft
feat(prefix-cache): sticky tools pins + DiffPin for warm agent TTFT
#567 opened Jul 31, 2026 by davidmroth Contributor Loading…
5 of 7 tasks
11
fix(ds4): make 16K graph eviction safe and parallelize index scan
#565 opened Jul 31, 2026 by davide221 Contributor Loading…
formal: add local prefix-cache contract registry
#564 opened Jul 31, 2026 by dusterbloom Collaborator • Draft
fix(prefix-cache): reuse slots after aborted snapshots
#563 opened Jul 31, 2026 by dusterbloom Collaborator Loading…
3
fix(draft): embed Qwen3.6 SWA metadata in Q8 converter
#562 opened Jul 30, 2026 by Amidwestnoob Contributor Loading…
feat(moe): add heterogeneous NVMe expert streaming tier
#560 opened Jul 30, 2026 by davide221 Contributor • Draft
2
build(deps): bump the github-actions group across 1 directory with 8 updates
#559 opened Jul 29, 2026 by dependabot Bot Loading…
1
feat: add paged-attention monolithic decode-only AR foundation
#555 opened Jul 23, 2026 by Graffioh Contributor Loading…
10
feat(ds4): add default-off HIP GPU phase profiler
#554 opened Jul 21, 2026 by cheese-cakee Contributor Loading…
11
fix(ds4): preserve exact speculative verification
#548 opened Jul 20, 2026 by cheese-cakee Contributor Loading…
3
feat(ds4): opt-in GPU F16 HC fn mirrors for ROCmFPX
#546 opened Jul 19, 2026 by Graffioh Contributor • Draft
feat(server): harden HTTP exposure and admission
#544 opened Jul 18, 2026 by da-beda • Draft
fix(server): make speculative decoding contract explicit
#543 opened Jul 18, 2026 by da-beda • Draft
fix(cmake): derive conservative minimum CUDA architecture
#542 opened Jul 18, 2026 by da-beda • Draft
bench(harness): add fail-closed validation v2 tooling
#541 opened Jul 18, 2026 by da-beda • Draft
bench(ds4): compare PRs with benchmarks
#535 opened Jul 16, 2026 by Graffioh Contributor • Draft
feat: DFlash support for tensor-parallel Qwen3.6 targets
#533 opened Jul 16, 2026 by jkyamog Contributor Loading…
28
perf(ds4): compute greedy argmax on GPU (monolithic and layer split)
#531 opened Jul 16, 2026 by Graffioh Contributor Loading…
4
refactor(draft): codify proposal row layout
#528 opened Jul 15, 2026 by dusterbloom Collaborator Loading…
2
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.