Ettore Di Giacinto c4058eb4da feat(paged): tail-fusion (0042) + full-step decode CUDA graph default-on (0043); FP4-MMA W4A4 (0034) + Marlin W4A16 (0035) MoE-GEMM scaffolds default-off ...

0042 fuses the pre-norm residual add into RMSNorm (+0.5% prefill, bit-exact). 0043 makes the full-step MoE decode CUDA graph default-on (+2-4% decode, bit-exact; removes ~18x per-step host kernel re-issue, A/B-confirmed). 0034 (native FP4-MMA W4A4) and 0035 (Marlin-style W4A16 grouped MoE GEMM) are correct + bit-exact but regress vs the int8 FP4-MMQ in-backend on GB10 (bf16 MMA is ~half the int8 rate); shipped default-off as validated mechanisms and recorded negatives per the parity methodology.

Assisted-by: Claude:opus-4.8 [Claude Code]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

2026-06-29 06:15:10 +00:00

..

ds4

chore: ⬆️ Update antirez/ds4 to 80ebbc396aee40eedc1d829222f3362d10fa4c6c (#10378)

2026-06-18 00:32:13 +02:00

grpc

fix: speedup git submodule update with --single-branch (#2847)

2024-07-13 22:32:25 +02:00

ik-llama-cpp

fix(ik-llama): port multimodal path to mtmd API and bump to f96eaddb (#10534) (#10568)

2026-06-28 08:57:11 +02:00

llama-cpp

paged: drop bf16-tau (patch 0026), subsumed by decode fusions (tau=100000 flat, zero speed benefit)

2026-06-28 16:06:06 +00:00

llama-cpp-localai-paged

feat(paged): tail-fusion (0042) + full-step decode CUDA graph default-on (0043); FP4-MMA W4A4 (0034) + Marlin W4A16 (0035) MoE-GEMM scaffolds default-off

2026-06-29 06:15:10 +00:00

privacy-filter

feat(backends): darwin/Metal build for the privacy-filter backend (#10513)

2026-06-26 01:18:41 +02:00

turboquant

fix(backends): quote $CURDIR in run.sh (fixes backends in paths with spaces) (#10519)

2026-06-26 01:02:48 +02:00

run-unit-tests.sh

fix(llama-cpp): stop reinterpreting plain-string message content as JSON (#10524) (#10538)

2026-06-27 01:42:05 +02:00