Repository navigation

merged master fd693ac6a2ab into apg - #7

Merged

stduhpf merged 83 commits into

stduhpf:apgfrom

wbruna:apg_merge_master

Oct 23, 2025

Merged

merged master fd693ac6a2ab into apg#7
stduhpf merged 83 commits into
stduhpf:apgfrom
wbruna:apg_merge_master

Commits on Jul 23, 2025
Commits on Jul 24, 2025
- docs: add LocalAI to README's UIs (#741)
  mudler
  authored
  fb86bf4
- sync: update ggml
  leejet
  committed
  eed97a5
Commits on Jul 28, 2025
- sync: update ggml
  leejet
  committed
  f54524f
- feat: upgrade musa sdk to rc4.2.0 (#732)
  yeahdongcn
  authored
  8c3c788
- feat: change image dimensions requirement for DiT models (#742)
  stduhpf
  authored
  59080d3
- feat: add missing models and parameters to image metadata (#743)
  
  wbruna
  and
  one-lithe-rune
  authored
  7eb30d0
- refector: optimize the usage of tensor_types
  leejet
  committed
  f6b9aa1
Commits on Aug 2, 2025
- feat: support build against system installed GGML library (#749)
  Seas0
  authored
  6167e29
- chore: avoid setting GGML_MAX_NAME when building against external ggml (#751)
  wbruna
  authored
  f7f05fb
- Conv2D direct support (#744)
  
  daniandtheweb
  and
  leejet
  authored
  5b8996f
- sync: update ggml, make cuda im2col a little faster
  leejet
  committed
  5900ef6
Commits on Sep 1, 2025
- chore: add Nvidia 30 series (cuda arch 86) to build
  nekopendev
  authored
  0e86d90
- feat: throttle model loading progress updates (#782)
  wbruna
  authored
  eea77cb
- docs: add missing dash to docs/chroma.md (#771)
  SmallAndSoft
  authored
  a7c7905
- docs: add compile option needed by Ninja (#770)
  SmallAndSoft
  authored
  f0fa7dd
- feat: show usage on unknown arg (#767)
  stduhpf
  authored
  4c6475f
Commits on Sep 3, 2025
- fix: typo in the verbose long flag (#783)
  wbruna
  authored
  2eb3845
Commits on Sep 6, 2025
- feat: add wan2.1/2.2 support (#778)
  leejet
  authored
  cb1d975
- feat: add support for timestep boundary based automatic expert routing in Wan MoE (#779)
  
  stduhpf
  and
  leejet
  authored
  21ce9fe
- feat: add flow shift parameter (for SD3 and Wan) (#780)
  
  stduhpf
  and
  leejet
  authored
  141a4b4
- docs: update docs and help message
  leejet
  committed
  d7f430c
Commits on Sep 7, 2025
- chore: update to c++17
  leejet
  committed
  675208d
- docs: update docs/wan.md
  leejet
  committed
  1c07fb6
- fix: add flash attn support check (#803)
  leejet
  authored
  f8fe4e7
- feat: support incrementing ref image index (omni-kontext) (#755)
  
  stduhpf
  and
  leejet
  authored
  c587a43
- feat: add detailed tensor loading time stat (#793)
  leejet
  authored
  c648001
Commits on Sep 8, 2025
- fix: clarify lora quant support and small fixes (#792)
  hartmark
  authored
  abb115c
Commits on Sep 10, 2025
- fix: accept NULL in sd_img_gen_params_t::input_id_images_path (#809)
  wbruna
  authored
  ff4fdbb
- chore: update flash attention warnings (#805)
  wbruna
  authored
  abb36d6
- fix: use {} for params init instead of memset (#781)
  wbruna
  authored
  ac5a215
- chore: remove sd3 flash attention warn (#812)
  leejet
  authored
  b017918
Commits on Sep 11, 2025
- feat: use log_printf to print ggml logs (#545)
  clibdev
  authored
  87cdbd5
- chore: add install() support in CMakeLists.txt (#540)
  clibdev
  authored
  6bbaf16
- feat: add SmoothStep Scheduler (#813)
  Green-Sky
  authored
  49d6570
- feat: add sd3 flash attn support (#815)
  leejet
  authored
  fce6afc
Commits on Sep 14, 2025
- fix: make tiled VAE reuse the compute buffer (#821)
  wbruna
  authored
  ddc4a18
- feat: reduce CLIP memory usage with no embeddings (#768)
  wbruna
  authored
  48956ff
- fix: make weight override more robust against ggml changes (#760)
  wbruna
  authored
  5869987
- fix: do not force VAE type to f32 on SDXL (#716)
  wbruna
  authored
  b54bec3
- feat: use Euler sampling by default for SD3 and Flux (#753)
  wbruna
  authored
  c607fc3
- fix: harden for large files (#643)
  Green-Sky
  authored
  9e7befa
- feat: Add SYCL Dockerfile (#651)
  richiejp
  authored
  a6a8569
- feat: increase work_ctx memory buffer size (#814)
  leejet
  authored
  dc46993
- docs: update docs
  leejet
  committed
  288e2d6
- feat: add VAE encoding tiling support and adaptive overlap (#484)
  
  authored
  2c9b1e2
- feat: add vace support (#819)
  leejet
  authored
  52a97b3
- feat: optimize tensor loading time (#790)
  
  rmatif
  and
  leejet
  authored
  55c2e05
- refactor: simplify the logic of pm id image loading (#827)
  leejet
  authored
  0ebe6fe
Commits on Sep 16, 2025
Commits on Sep 17, 2025
- chore: fix workflow (#836)
  leejet
  authored
  1e5f207
- fix: avoid multithreading issues in the model loader
  leejet
  committed
  567f9f1
- fix: avoid segfault for pix2pix models without reference images (#766)
  
  wbruna
  and
  leejet
  authored
  171b222
- refactor: remove unused --normalize-input parameter (#835)
  leejet
  authored
  fd693ac
Commits on Sep 20, 2025
- merge with master fd693ac6a2ab
  wbruna
  committed
  057abde
Commits on Sep 24, 2025
- fix: correct tensor deduplication logic (#844)
  rmatif
  authored
  1e0d282
- docs: include Vulkan compatibility for LoRA quants (#845)
  wbruna
  authored
  513f36d
- docs: HipBLAS / ROCm build instruction fix (#843)
  Stefan-Olt
  authored
  98ba155
- fix: tensor loading thread count (#854)
  wbruna
  authored
  f3140ea
- fix: optimize the handling of CLIP embedding weight (#840)
  leejet
  authored
  2abe945
- sync: update ggml
  leejet
  committed
  1ba30ce
Commits on Sep 25, 2025
- sync: update ggml
  leejet
  committed
  6ad46bb
- fix: optimize the handling of embedding weight (#859)
  leejet
  authored
  35843c7
Commits on Oct 10, 2025
- feat: add support for Flux Controls and Flex.2 (#692)
  stduhpf
  authored
  11f436c
- docs: update README.md (#866)
  sharuzzaman
  authored
  b451728
- chore: fix dockerfile libgomp1 dependency + improvements (#852)
  SergeantSerk
  authored
  940a201
- fix: ensure directory iteration results are sorted by filename (#858)
  leejet
  authored
  e12d5e0
- chore: fix vulkan ci (#878)
  leejet
  authored
  02af48a
Commits on Oct 12, 2025
- feat: add support for more esrgan models & x2 & x1 models (#855)
  pedroCabrera
  authored
  e70d020
- feat: add a stand-alone upscale mode (#865)
  
  wbruna
  and
  leejet
  authored
  5b261b9
- refactor: deal with default img-cfg-scale at the library level (#869)
  wbruna
  authored
  aa68b87
- feat: add Qwen Image support (#851)
  leejet
  authored
  beb99a2
- fix: resolve VAE tiling problem in Qwen Image (#873)
  wbruna
  authored
  9727c6b
- fix: avoid generating black images when running T5 on the GPU (#882)
  leejet
  authored
  1c32fa0
Commits on Oct 13, 2025
- fix: correct canny preprocessor (#861)
  wbruna
  authored
  5436f6b
- fix: better progress display for second-order samplers (#834)
  wbruna
  authored
  c64994d
- feat: add Qwen Image Edit support (#877)
  leejet
  authored
  2e9242e
- Merge 'master-323-2e9242e' into apg_merge_master
  wbruna
  committed
  4fa6350

Commits

Commits on Jul 23, 2025

Commits on Jul 24, 2025

Commits on Jul 28, 2025

Commits on Aug 2, 2025

Commits on Sep 1, 2025

Commits on Sep 3, 2025

Commits on Sep 6, 2025

Commits on Sep 7, 2025

Commits on Sep 8, 2025

Commits on Sep 10, 2025

Commits on Sep 11, 2025

Commits on Sep 14, 2025

Commits on Sep 16, 2025

Commits on Sep 17, 2025

Commits on Sep 20, 2025

Commits on Sep 24, 2025

Commits on Sep 25, 2025

Commits on Oct 10, 2025

Commits on Oct 12, 2025

Commits on Oct 13, 2025