Repository navigation

Resync with upstream master-1050 and port the ggml patches - #52

Open

danielhanchen wants to merge 72 commits into

masterfrom

resync/upstream-master-1050

Open

Resync with upstream master-1050 and port the ggml patches#52
danielhanchen wants to merge 72 commits into
masterfrom
resync/upstream-master-1050

Commits on Oct 21, 2026
Commits on Oct 22, 2026
Commits on Oct 23, 2026
- fix: forward ref_image_args to the video pipeline (#2177)
  philipsorst
  authored
  baf45b4
- refactor: centralize model factories and defaults (#2198)
  leejet
  authored
  cd7f986
- feat: allow grouped temporal decoding for Wan2.2 TI2V 5B (#2196)
  losewayy
  authored
  40b1788
- chore: format code
  leejet
  committed
  e0ed724
- fix: update ggml to fix SageAttention builds on older GPUs (#2199)
  leejet
  authored
  95ff762
Commits on Oct 24, 2026
Commits on Oct 25, 2026
- feat: print sigma values before inference (#2205)
  wbruna
  authored
  d234ac6
- perf: reuse shared modulation slices for moss image (#2208)
  mirek190
  authored
  2a5f810
- perf: avoid redundant MOSS QKV and value layout copies (#2209)
  
  mirek190
  and
  leejet
  authored
  e6370bf
- feat: support FLUX.3 Diffusers LoRA tensor names (#2214)
  leejet
  authored
  bb6a0cc
- feat(server): accept ref_videos, ref_audios, ref_video_audios in sdcpp vid_gen (#2194)
  crystalsighting
  authored
  0f76bf5
- sync: update frontend
  leejet
  committed
  b29962d
- chore: reduce large recipe defaults to 1024x1024 except PiD
  leejet
  committed
  a5e48c6
- ci: trigger builds on recipe JSON changes
  leejet
  committed
  012c7f8
- feat(server): add recipe constraints and update generation workspace (#2217)
  leejet
  authored
  e41fbcf
Commits on Oct 26, 2026
- fix: avoid repeated seeks when skipping small GGUF metadata values (#2223)
  leejet
  authored
  bb6b128
- fix: enable segmented execution for FLUX.3 (#2224)
  leejet
  authored
  be48649
Commits on Oct 27, 2026
Commits on Oct 28, 2026
- refactor: use a distinct version id for Chroma (#2237)
  wbruna
  authored
  8cbe905
- perf: accelerate INT8 diffusion inference with CUDA fusions (#2235)
  leejet
  authored
  77b72a2
Commits on Oct 29, 2026
Commits on Oct 31, 2026
- perf: optimize Vulkan kernels, graph fusion, and workspace reuse (#2251)
  leejet
  authored
  0a06712
- perf(vulkan): update ggml for H3 inference optimizations (#2252)
  leejet
  authored
  c6dfa9d
Commits on Nov 1, 2026
Commits on Nov 2, 2026
- ci: cache CUDA compilation for Windows and Docker builds
  leejet
  committed
  1350cf5
- fix: layer split across cuda (#2262)
  cumal
  authored
  250ea6e
- fix: support integer LoRA scalar conversion to FP32 (#2263)
  leejet
  authored
  b4d6b83
- ci: use ccache to fix Windows CUDA command parsing failures
  leejet
  committed
  3be2378
Commits on Nov 3, 2026
Commits on Nov 4, 2026
- feat: add P-H3-Image generation and editing support (#2272)
  leejet
  authored
  3e92bf9
- feat: MiniMax-H3 latent hires second pass (#2249)
  farawayso
  authored
  d135894
- perf(vulkan): accelerate MOSS normalization and SiTU staging (#2273)
  
  leejet
  and
  mirek190
  authored
  58bb3c1
Commits on Nov 6, 2026

Commits

Commits on Oct 21, 2026

Commits on Oct 22, 2026

Commits on Oct 23, 2026

Commits on Oct 24, 2026

Commits on Oct 25, 2026

Commits on Oct 26, 2026

Commits on Oct 27, 2026

Commits on Oct 28, 2026

Commits on Oct 29, 2026

Commits on Oct 31, 2026

Commits on Nov 1, 2026

Commits on Nov 2, 2026

Commits on Nov 3, 2026

Commits on Nov 4, 2026

Commits on Nov 6, 2026