Skip to content

Pull requests: Luce-Org/lucebox

New pull request New
74 Open 533 Closed
74 Open 533 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

refactor(server): establish owned generation lifecycle
#692 opened Sep 2, 2026 by Graffioh Contributor • Draft
7
refactor(server): seal backend construction behind a plan
#691 opened Sep 2, 2026 by Graffioh Contributor Loading…
5
fix(server): unify generation failure responses
#689 opened Sep 1, 2026 by Graffioh Contributor Loading…
5
feat(test): add kernel qualification coverage for GDN, MoE, and DeepSeek4
#687 opened Sep 1, 2026 by Graffioh Contributor Loading…
13
perf(qwen35): retain speculation during paged prefill
#686 opened Aug 31, 2026 by Graffioh Contributor • Draft
perf(ds4): unify gfx1151 sparse-prefill stack
#685 opened Aug 31, 2026 by davide221 Contributor • Draft
1
perf(ds4): extend sparse prefill scheduling to 10K
#684 opened Aug 30, 2026 by dusterbloom Collaborator • Draft
1
perf(ds4): default mixed ROCmFP MMQ prefill on gfx1151
#683 opened Aug 30, 2026 by davide221 Contributor • Draft
7
fix(pflash): score the user query and reject non-finite scores
#679 opened Aug 28, 2026 by Graffioh Contributor • Draft
fix(ci) make manual speed profiles trustworthy
#678 opened Aug 28, 2026 by davide221 Contributor Loading…
2
fix(ggml-cuda): require nonzero graph ID for property bypass
#676 opened Aug 27, 2026 by cheese-cakee Contributor Loading…
3
feat(pflash): improve AMD defaults and enable rocWMMA
#669 opened Aug 27, 2026 by Graffioh Contributor • Draft
perf(ds4): accelerate ROCm decode and long-context selection
#667 opened Aug 26, 2026 by davide221 Contributor Loading…
20
perf(ds4): batch exact prefill through width-4 vector kernels
#666 opened Aug 26, 2026 by dusterbloom Collaborator Loading…
3
fix(server): prefix-cache deepening (stop pinning the restore point + stop truncating at stray end-markers)
#665 opened Aug 26, 2026 by jkyamog Contributor Loading…
11
perf(ds4): accelerate long context and add PFlash
#664 opened Aug 26, 2026 by davide221 Contributor • Draft
5
experimental: integrate luce_odistill bridge
#663 opened Aug 26, 2026 by Graffioh Contributor • Draft
perf(ds4): fuse expert-major MoE route combine
#658 opened Aug 25, 2026 by cheese-cakee Contributor Loading…
11
feat(server): add prefix cache to paged concurrency (qwen AR)
#655 opened Aug 24, 2026 by Graffioh Contributor Loading…
18
feat(test): add manual Qwen 3.6 production qualification
#636 opened Aug 20, 2026 by Graffioh Contributor Loading…
102
perf(ds4): elide intermediate prefill logits
#633 opened Aug 18, 2026 by cheese-cakee Contributor • Draft
10
fix(qwen35): size mixed concurrent target-step graphs
#622 opened Aug 18, 2026 by Graffioh Contributor • Draft
feat(harness): add interactive modes and GPU selection
#611 opened Aug 15, 2026 by Graffioh Contributor Loading…
2 tasks
5
feat(mix): batched (MMQ) path for qtypes 105/106, routed on batch width
#607 opened Aug 13, 2026 by geometric Bot Loading…
1
feat(harness): add OMP client support
#603 opened Aug 12, 2026 by Graffioh Contributor Loading…
2 tasks
16
ProTip! Add no:assignee to see everything that’s not assigned.