Skip to content

Pull requests: sonos/tract

New pull request New
35 Open 1,994 Closed
35 Open 1,994 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(causal_llm): add LogitsProcessor hook for guidance / constrained decoding
#2381 opened Jun 16, 2026 by JulienBalianSonos Collaborator Loading…
release.sh/post-release.sh: author the release and post-release commi…
#2380 opened Jun 16, 2026 by kali Collaborator Loading…
Feat/bench pr
#2379 opened Jun 16, 2026 by kali Collaborator Loading…
1
examples/causal_llm: speculative decoding (n-gram + draft-model)
#2370 opened Jun 14, 2026 by czoli1976 Contributor Loading…
metal/ggml: keep q4_0 decode on the mat-vec kernel up to 8 rows
#2369 opened Jun 14, 2026 by czoli1976 Contributor Loading…
linalg,core: SIMD ReduceMin (mirror the max reducer)
#2368 opened Jun 13, 2026 by czoli1976 Contributor Loading…
core: don't recompute ReduceMax scalar after the SIMD max kernel
#2367 opened Jun 13, 2026 by czoli1976 Contributor Loading…
metal: f16 activations/output for the GGML matmul (drop f32 round-trips)
#2366 opened Jun 13, 2026 by czoli1976 Contributor Loading…
metal: pack element-wise kernels into full-width threadgroups
#2365 opened Jun 13, 2026 by czoli1976 Contributor Loading…
core: drop per-run String/TypedFact alloc in resolve_symbols_with_states
#2364 opened Jun 13, 2026 by czoli1976 Contributor Loading…
move clean Resize subset and GridSample into tract-core
#2363 opened Jun 12, 2026 by JulienBalianSonos Collaborator Loading…
linalg/cache sorcery: LLC/SLC-aware budget for the L3 outer blocking tier
#2352 opened Jun 7, 2026 by czoli1976 Contributor Loading…
1
linalg: add Q1_58 ternary (BitNet b1.58) block-quant format
#2351 opened Jun 7, 2026 by czoli1976 Contributor Loading…
linalg/mmm: L3-resident outer tier for the single-thread block walk
#2350 opened Jun 7, 2026 by czoli1976 Contributor Loading…
3
linalg: reusable CPU cache-size detection module
#2349 opened Jun 7, 2026 by czoli1976 Contributor Loading…
linalg/block_quant: W4A8 int8-dot decode GEMV for Q4_0
#2348 opened Jun 7, 2026 by czoli1976 Contributor Loading…
onnx: run MatMulNBits int4 through the fused Q4_0 block-quant matmul
#2340 opened Jun 4, 2026 by czoli1976 Contributor Loading…
1
linalg/x86_64: Intel AMX and AVX-VNNI int8/bf16 GEMM kernels
#2339 opened Jun 3, 2026 by czoli1976 Contributor Loading…
4
transformers: KIVI-style KV-cache quantization — packed u8 storage, ~4× memory vs f32
#2329 opened Jun 2, 2026 by czoli1976 Contributor Loading…
2
transformers: in-place KV cache for decode via a fused InPlaceKvSdpa op
#2321 opened May 31, 2026 by czoli1976 Contributor Loading…
1
transformers: route fused ScaledMaskedSoftmax through an accurate vectorized exp
#2318 opened May 29, 2026 by czoli1976 Contributor Loading…
2
linalg/arm64/sve: VLA SVE2 rms_norm_f32 kernel (stacked on #2314)
#2315 opened May 29, 2026 by czoli1976 Contributor Loading…
5
linalg/arm64/sme: experimental f16 SME kernels for Apple M5/A19 Need tester with hardware
#2273 opened May 22, 2026 by czoli1976 Contributor • Draft
feat: add tract_moe_ffn operator for Mixture-of-Experts FFN
#2084 opened Mar 30, 2026 by JulienBalianSonos Collaborator Loading…
Flux schnell
#2082 opened Mar 30, 2026 by kali Collaborator Loading…
ProTip! Add no:assignee to see everything that’s not assigned.