Skip to content

v2.3.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 25 Mar 05:09
· 35 commits to main since this release
v2.3.0
22a82bf

What's New in v2.3.0

The March 2026 revival — 90+ commits rebuilding the system from the ground up.

Desktop GUI

  • Tkinter application with Sun Valley dark/light theme
  • Two-tab layout: Search & Index + Web Research
  • LLM streaming with live token rendering and collapsible thinking blocks
  • RenderedMarkdown widget with syntax-highlighted code, tables, clickable links
  • Preferences dialog with endpoint presets and Test Connection

Web Research Engine

  • rag-mini scrape — fetch URLs and extract clean markdown (HTML, PDF, arXiv, GitHub)
  • rag-mini search-web — search DuckDuckGo, Tavily, or Brave
  • rag-mini research — full pipeline: search → scrape → index
  • rag-mini research --deep — iterative ANALYZE→SEARCH→SCRAPE→PRUNE→REPORT cycles with time budgets

Search Architecture Rebuild

  • Independent dual-pipeline search: semantic + BM25 run against the full index separately
  • Reciprocal Rank Fusion (RRF) merges results by rank position
  • Code-aware BM25 tokenizer splits snake_case and CamelCase
  • Smart re-ranking, diversity constraints, chunk consolidation
  • Auto-calibrating score labels (HIGH/GOOD/FAIR/LOW)

Embedding System

  • OpenAI-compatible endpoint support (LM Studio, vLLM, OpenAI, any proxy)
  • Auto-detection via /v1/models with two profiles: precision (MiniLM) and conceptual (Nomic)
  • No fake fallbacks — if nothing is available, BM25 still works

Corpus Management

  • Vector-based deduplication using indexed embeddings (cosine similarity)
  • LLM borderline classification for 0.80-0.95 similarity range
  • Cross-document corroboration detection
  • Rate limiting with exponential backoff for all API calls

Downloads

Platform File Notes
Windows fss-mini-rag-2.3.0-setup.exe Standalone installer — no Python needed
Linux (Debian/Ubuntu) fss-mini-rag_2.3.0_amd64.deb sudo dpkg -i *.deb
Linux (Portable) FSS-Mini-RAG-2.3.0-x86_64.AppImage Single file, chmod +x and run
Any (pip) fss_mini_rag-2.3.0-py3-none-any.whl pip install *.whl
Any (portable) rag-mini.pyz python rag-mini.pyz (needs Python)

One-Line Install (Linux/macOS)

curl -fsSL https://raw.githubusercontent.com/FSSCoding/Fss-Mini-Rag/main/install.sh | bash

From Source

git clone https://github.com/FSSCoding/Fss-Mini-Rag.git
cd Fss-Mini-Rag
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt && pip install -e .
rag-mini gui

Changes

  • Fix AppImage build: create dist/ directory before appimagetool
Assets 8
Loading