mirrors/Open-Generative-AI

Fork 0

mirror of https://github.com/Anil-matcha/Open-Generative-AI.git synced 2026-05-07 01:17:18 +00:00

Commit graph

Author	SHA1	Message	Date
Anil Matcha	0493e6244b	fix(local-ai): walk back releases when latest is partial; broaden Linux match Fixes #108 — leejet's latest sd.cpp release (master-587-b8bdffc) only ships Mac arm64, the Windows cudart runtime stub, and Linux ROCm. The old matcher only ever read `releases/latest` and excluded ROCm/Vulkan, so Ubuntu/Linux users hit "No binary found for this platform" with no recovery path. Windows users were also broken: the only Win asset in the latest release isn't an sd-cli build. - Walk the last 15 leejet releases until one ships a usable build for the current platform, so a partial latest release self-heals to the prior tag (master-586 onwards has the full 12-zip matrix). - Linux: prefer plain x86_64, then vulkan, then rocm (instead of rejecting both rocm and vulkan). - Windows: priority order avx2 > avx > avx512 > noavx > cuda12, and skip the standalone `cudart-sd-bin-win-cu12-x64.zip` runtime stub. - macOS Intel: surface a clear error — leejet only ships arm64. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-25 10:51:44 +05:30
Anil Matcha	6011fcb0fd	feat: use Metal-GPU binary for macOS arm64 local inference macOS Apple Silicon (darwin-arm64) now downloads the Metal-accelerated sd-cli from our own GitHub release instead of the stock leejet build, enabling significantly faster local image generation on M1/M2/M3/M4. All other platforms continue to use the leejet upstream release. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-22 18:02:33 +05:30
Anil Matcha	36d392ab78	feat: add local model inference via sd.cpp (Electron-only) - Scaffold full IPC bridge: preload.js exposes localAI API to renderer - electron/lib/localInference.js: binary download (Metal-enabled macOS build), model management, auxiliary file downloads (Qwen3-4B LLM + FLUX VAE for Z-Image), and generation via sd-cli with DYLD_LIBRARY_PATH + xattr quarantine stripping - electron/lib/modelCatalog.js: Z-Image Turbo/Base (featured) + SD 1.5/SDXL models - Z-Image requires 3 components loaded via --diffusion-model + --llm + --vae flags - ImageStudio: Local/API toggle (Electron-only), local model selector, progress bar - SettingsModal: tabbed UI with Local Models tab (hidden on web/hosted version) - LocalModelManager: engine status bar, per-model download cards, auxiliary component download UI (text encoder + VAE) for Z-Image models - localInferenceClient.js: safe wrapper with isLocalAIAvailable() guard - All local features gated behind isLocalAIAvailable() — web version unaffected Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-22 17:47:17 +05:30

Author

SHA1

Message

Date

Anil Matcha

0493e6244b

fix(local-ai): walk back releases when latest is partial; broaden Linux match

Fixes #108 — leejet's latest sd.cpp release (master-587-b8bdffc) only
ships Mac arm64, the Windows cudart runtime stub, and Linux ROCm. The
old matcher only ever read `releases/latest` and excluded ROCm/Vulkan,
so Ubuntu/Linux users hit "No binary found for this platform" with no
recovery path. Windows users were also broken: the only Win asset in
the latest release isn't an sd-cli build.

- Walk the last 15 leejet releases until one ships a usable build for
  the current platform, so a partial latest release self-heals to the
  prior tag (master-586 onwards has the full 12-zip matrix).
- Linux: prefer plain x86_64, then vulkan, then rocm (instead of
  rejecting both rocm and vulkan).
- Windows: priority order avx2 > avx > avx512 > noavx > cuda12, and
  skip the standalone `cudart-sd-bin-win-cu12-x64.zip` runtime stub.
- macOS Intel: surface a clear error — leejet only ships arm64.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-04-25 10:51:44 +05:30

Anil Matcha

6011fcb0fd

feat: use Metal-GPU binary for macOS arm64 local inference

macOS Apple Silicon (darwin-arm64) now downloads the Metal-accelerated
sd-cli from our own GitHub release instead of the stock leejet build,
enabling significantly faster local image generation on M1/M2/M3/M4.
All other platforms continue to use the leejet upstream release.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-04-22 18:02:33 +05:30

Anil Matcha

36d392ab78

feat: add local model inference via sd.cpp (Electron-only)

- Scaffold full IPC bridge: preload.js exposes localAI API to renderer
- electron/lib/localInference.js: binary download (Metal-enabled macOS build),
  model management, auxiliary file downloads (Qwen3-4B LLM + FLUX VAE for Z-Image),
  and generation via sd-cli with DYLD_LIBRARY_PATH + xattr quarantine stripping
- electron/lib/modelCatalog.js: Z-Image Turbo/Base (featured) + SD 1.5/SDXL models
- Z-Image requires 3 components loaded via --diffusion-model + --llm + --vae flags
- ImageStudio: Local/API toggle (Electron-only), local model selector, progress bar
- SettingsModal: tabbed UI with Local Models tab (hidden on web/hosted version)
- LocalModelManager: engine status bar, per-model download cards, auxiliary
  component download UI (text encoder + VAE) for Z-Image models
- localInferenceClient.js: safe wrapper with isLocalAIAvailable() guard
- All local features gated behind isLocalAIAvailable() — web version unaffected

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-04-22 17:47:17 +05:30

3 commits