Skip to content

v1.1.0

Choose a tag to compare

View all tags
@inureyes inureyes released this 10 Feb 01:01
· 1 commit to main since this release
v1.1.0
f35b1a0

Draw page for conversation-style image generation, Stable Diffusion integration, document translation with glossary, dynamic context length presets, and headless server mode.

Backend.AI GO v1.1.0

This is the first feature release since v1.0.2, consolidating all changes from the beta cycle (beta.1 through beta.3) into a stable release. This release includes 117 commits with major new features, improvements, and bug fixes.

New Features

  • Draw Page — Conversation-style image generation interface with dual generation flow supporting both local and cloud models (,)
  • Stable Diffusion Integration — sd-server backend for image generation with DiffusionConfigDrawer for server parameters, engine installation prompts, and auto-save to Creations (,,)
  • Document Translation — Support for TXT, MD, DOCX, PDF, and image OCR translation with glossary management
  • Dynamic Context Length — Model-aware context length presets with automatic adjustment
  • Headless Server Mode — Standalone bgo-server binary with embedded frontend for server deployments
  • Clipboard Image Paste — Paste images directly into chat message input
  • Help Menu — Documentation links and Acknowledgments in the app menu bar
  • Configurable Health Check — Per-backend health check endpoint override for OpenAI-compatible providers (,)
  • Bookmark Responses — Bookmark button on chat responses with Image/Text tabs in Creations page
  • Use Case Guides — Documentation for building apps with the API and private document translation
  • Common ModelSelect — Reusable model selector component for Draw page
  • SBOM Generation — Software Bill of Materials in CI packaging pipeline
  • Acknowledgments — Link to About modal with SBOM references

Improvements

  • Persist and restore Draw session images from disk
  • Conditional parameter panels based on model type in Draw page
  • Fetch router models for Draw page model selector
  • Tooltip for negative prompt toggle explaining model-specific support
  • Unified sidebar menu names using nouns consistently
  • Unified ProcessManager and SdProcessManager with trait-based architecture

Bug Fixes

  • Use backendType to detect diffusion models in Canvas model selector
  • Make Draw parameter panel collapse visible with sliding toggle button
  • Clamp ModelSelect dropdown position to stay within viewport
  • Prevent onboarding tour from reappearing after completion
  • Prevent Agent tab hang after tool denial and LLM API 500 error
  • Update Help menu URLs to correct repositories and localized docs
  • Prevent unnecessary re-inference when navigating back to Translation tab
  • Update sd-cpp upstream to fix Metal DIAG_MASK_INF crash
  • Use dedicated auto-detect prompt for translation
  • Register sd-server sidecar and emit error event on load failure
  • Handle empty level/source filters in log viewer
  • Use local state for API key and socket directory inputs
  • Use opaque background for ConfirmDialog
  • Emit 'complete' event for diffusion model loading
  • Add model loading progress bar for diffusion models
  • Enable Metal GPU acceleration for sd-cpp on macOS
  • Load conversation list on dashboard mount with limit support
  • Remove sd-server from externalBin to fix packaging CI

CI/CD

  • Generate SBOM in CI packaging pipeline

Technical Details

  • 117 commits since v1.0.2
  • Unified process management with trait-based architecture
  • Prettier formatting applied to Draw components and stores

Full Changelog

What's Changed

  • feat: Translation Feature - Phase 5: Document Translation (TXT/MD) by @inureyes
  • feat: DOCX document translation support by @inureyes
  • feat: Translation Feature - Phase 7: Document Translation (PDF) by @inureyes
  • fix: add missing i18n translations and technical reports by @inureyes
  • feat: Translation Feature - Phase 8: Image OCR Translation by @inureyes
  • fix: accumulate download chunk sizes for accurate progress display by @inureyes
  • Add random suffix to Unix socket paths to prevent collision during concurrent model loading by @inureyes
  • feat: unify file translation UI for images and documents by @inureyes
  • feat: add glossary feature for translation terminology management by @inureyes
  • fix: use package index version for runtime download URL by @inureyes
  • feat: improve translation glossary integration and add missing i18n by @inureyes
  • fix: add Open Graph and Twitter Card meta tags to root redirect page by @inureyes
  • refactor: resolve clippy too_many_arguments warnings by @inureyes
  • fix: replace default Next.js favicon with app icon by @inureyes
  • fix: update MkDocs favicon and logo to match current app icon by @inureyes
  • feat: increase default context length from 4096 to 8192 tokens by @inureyes
  • feat: add dynamic context length based on model metadata by @inureyes
  • feat: add context length presets based on usage patterns by @inureyes
  • fix: unify glossary page design and show full language names by @inureyes
  • update: cargo dependencies to latest compatible versions by @inureyes
  • docs: add OG/Twitter meta tags and per-page descriptions to MkDocs by @inureyes
  • fix: update X(Twitter) handle to @lablupinc by @inureyes
  • chore: update Tauri NPM packages to match Rust crate versions by @inureyes
  • docs: add use case guide for using Claude Code with Backend.AI GO by @inureyes
  • fix: correct TCP server location in Claude Code use case docs by @inureyes
  • feat: add configurable default max tokens setting with 32K default by @inureyes
  • fix: enable text selection in chat message bubbles by @inureyes
  • fix: resolve cursor-jump and add sliders for number inputs in settings by @inureyes
  • feat: add widget drag-to-resize and improve dashboard grid layout by @inureyes
  • fix: prevent getting started banner flash on app restart by @inureyes
  • fix: prevent default theme flash on app startup by @inureyes
  • fix: apply glossary as source text pre-processing for TranslateGemma by @inureyes
  • docs: add PDF manual generation with mkdocs-exporter by @inureyes
  • fix: use configurable health endpoint path for sd-server by @inureyes
  • docs: add coming soon note to iOS installation section by @inureyes
  • feat: add single-file Stable Diffusion models to Getting Started featured models by @inureyes
  • fix: tolerate SHA256 hash mismatch when file size matches expected by @inureyes
  • feat: integrate sd-server for diffusion model loading and image generation by @inureyes
  • feat: prompt engine installation for diffusion models by @inureyes
  • fix: load conversation list on dashboard mount with limit support by @inureyes
  • docs: add use case guide for private document translation by @inureyes
  • feat: add bookmark button on chat responses and Image/Text tabs in Creations page by @inureyes
  • fix: onboarding tour fixes and keyboard shortcuts modal improvements by @inureyes
  • feat: show loaded diffusion models in model list and register with router by @inureyes
  • feat: add DiffusionConfigDrawer with stable-diffusion.cpp server parameters by @inureyes
  • feat: Add Draw page for conversation-style image generation (Phase 1) by @inureyes
  • docs: add use case guide - Building Apps with the API by @inureyes
  • feat: auto-save Draw page images to Creations by @inureyes
  • fix: add model loading progress bar for diffusion models (sd-cpp) by @inureyes
  • fix: emit 'complete' event for diffusion model loading by @inureyes
  • fix: use opaque background for ConfirmDialog by @inureyes
  • feat(draw): replace model selector with common ModelSelect component by @inureyes
  • fix: use local state for API key and socket directory inputs by @inureyes
  • feat(draw): add conditional parameter panels based on model type by @inureyes
  • feat: support clipboard image paste in chat message input by @inureyes
  • fix: handle empty level/source filters in log viewer by @inureyes
  • feat(draw): implement dual generation flow for local and cloud models by @inureyes
  • feat: Add Help menu with documentation links and Acknowledgments by @inureyes
  • fix: register sd-server sidecar and emit error event on load failure by @inureyes
  • fix: use dedicated auto-detect prompt for translation by @inureyes
  • fix: update sd-cpp upstream to fix Metal DIAG_MASK_INF crash by @inureyes
  • fix: prevent unnecessary re-inference when navigating back to Translation tab by @inureyes
  • fix: update Help menu URLs to correct repositories and localized docs by @inureyes
  • fix: prevent Agent tab hang after tool denial and LLM API 500 error by @inureyes
  • fix: prevent onboarding tour from reappearing after completion by @inureyes
  • feat: add Acknowledgments link to About modal and update SBOM references by @inureyes
  • feat: fetch router models for Draw page model selector by @inureyes
  • update: unify sidebar menu names to use nouns consistently by @inureyes
  • feat: generate SBOM in CI packaging pipeline by @inureyes
  • feat: add tooltip to negative prompt toggle explaining model-specific support by @inureyes
  • feat: persist and restore Draw session images from disk by @inureyes
  • fix: clamp ModelSelect dropdown position to stay within viewport by @inureyes
  • fix: make Draw parameter panel collapse visible with sliding toggle button by @inureyes
  • fix: use backendType to detect diffusion models in Canvas model selector by @inureyes
  • feat: add configurable health check endpoint for OpenAI-compatible providers by @inureyes
  • fix: add per-backend health check override for sd-server and show loaded diffusion models by @inureyes
  • refactor: unify ProcessManager and SdProcessManager with trait-based architecture by @inureyes

Contributors

inureyes
Assets 15
Loading