Skip to content

CHANGELOG.md

Latest commit

 

History

History
48 lines (40 loc) · 3.26 KB

CHANGELOG.md

File metadata and controls

48 lines (40 loc) · 3.26 KB
Raw

Changelog

v2.2 - 2026-05-28

  • Added a vLLM OpenAI compatible provider for local/Runpod models.
  • Added Settings fields for vLLM endpoint URL, model type, and API key.
  • Defaulted the vLLM model type to Qwen/Qwen3-8B and normalized Runpod base URLs to /v1/chat/completions.
  • Added a vLLM model preset dropdown for Qwen2.5-VL, Qwen3-VL, Qwen3.5, Gemma 3, Pixtral, and Mistral Small models.
  • Added vllm.md with Runpod and local multimodal vLLM setup notes.

v2.0 - 2026-05-10

  • Reworked Settings into dedicated Tagging Options and Captioning Options sections.
  • Added separate default models for tagging and captioning.
  • Moved max tags, max caption length, and editable system prompts into Settings instead of hiding them in the batch modals.
  • Added Crop Options with free-form mode, common aspect-ratio presets, and configurable crop output width.
  • Crops now use the selected aspect ratio and resize output from the saved crop width.
  • Replaced the single model/API-key picker with a provider API key table for Gemini, Grok, OpenAI, Claude, DeepDanbooru, and WD1.4.
  • Added visible key status per provider so saved-key state is obvious.
  • Hid the inline trigger-word and dataset-name labels while users are typing, because text overlap is not a feature.
  • Added a bottom-right Part of Lora Pilot family link to the LoRA Pilot GitHub repository.
  • Kept TagPilot as a single static tagpilot.html file with no install/build step. Miracles happen, occasionally.
  • Refactored provider calls into a provider registry.
  • Rendered tag pills without interpolating tag text into HTML.
  • Shared the batch processing runner between tagging and captioning.
  • Added object URL lifecycle cleanup for previews, crops, exports, removals, and resets.
  • Switched image-card actions to stable item IDs instead of shifting array indexes.
  • Extracted shared tag parsing, formatting, merge, and trigger-word helpers.
  • Expanded the no-install Node regression harness to cover settings, provider routing, refactor seams, and UI-state behavior.

v1.6 - 2026-05-08

  • Added Claude support through Anthropic's browser-direct Messages API using claude-sonnet-4-5-20250929.
  • Updated cloud provider defaults to current request shapes: Gemini 3.1 Flash Lite, Grok 4.3 via xAI Responses, and OpenAI GPT-5.4 Mini via Responses.
  • Wired custom tag and caption prompts into provider requests instead of leaving the UI knobs there for decoration.
  • Enforced the caption word limit after generation and included it in caption prompts.
  • Moved Settings to the top-left corner and added a visible Settings label next to the icon.
  • Fixed provider API key switching so Gemini, Claude, Grok, and OpenAI each reload their own saved key.
  • Added clearer provider HTTP errors, including useful OpenAI 401 messages.
  • Added a per-image processing indicator for single-image Tag and Caption actions.
  • Added a Crop button directly inside the image preview modal.
  • Added a no-install Node test harness for provider payloads and UI-state regressions.

v1.5

  • Added ZIP/image dataset loading with paired .txt captions.
  • Added trigger-word management, global tag cleanup, per-image tag/caption editing, duplicate detection, cropping, and ZIP export.
  • Added multiple tagging backends: Gemini, Grok, OpenAI, DeepDanbooru, and WD1.4.