BobAi
a84ff94fba
This comprehensive update enhances user experience with several key improvements: ## Enhanced Streaming & Thinking Display - Implement real-time streaming with gray thinking tokens that collapse after completion - Fix thinking token redisplay bug with proper content filtering - Add clear "AI Response:" headers to separate thinking from responses - Enable streaming by default for better user engagement - Keep thinking visible for exploration, collapse only for suggested questions ## Natural Conversation Responses - Convert clunky JSON exploration responses to natural, conversational format - Improve exploration prompts for friendly, colleague-style interactions - Update summary generation with better context handling - Eliminate double response display issues ## Model Reference Updates - Remove all llama3.2 references in favor of qwen3 models - Fix non-existent qwen3:3b references, replace with proper model names - Update model rankings to prioritize working qwen models across all components - Ensure consistent model recommendations in docs and examples ## Cross-Platform Icon Integration - Add desktop icon setup to Linux installer with .desktop entry - Add Windows shortcuts for desktop and Start Menu integration - Improve installer user experience with visual branding ## Configuration & Navigation Fixes - Fix "0" option in configuration menu to properly go back - Improve configuration menu user-friendliness - Update troubleshooting guides with correct model suggestions These changes significantly improve the beginner experience while maintaining technical accuracy and system reliability.
145 lines
10 KiB
YAML
145 lines
10 KiB
YAML
# FSS-Mini-RAG Configuration - Beginner-Friendly Edition
|
|
#
|
|
# 🎯 QUICK START PRESETS:
|
|
# - Keep defaults for most cases (recommended for beginners)
|
|
# - For large projects (>10k files): increase max_size to 3000
|
|
# - For faster search: set similarity_threshold to 0.2
|
|
# - For better results: enable expand_queries (but slower search)
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 📝 CHUNKING: How we break up your code files for searching
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# Think of chunks as "bite-sized pieces" of your code that the system can search through.
|
|
# Smaller chunks = more precise results but might miss context
|
|
# Larger chunks = more context but might be less precise
|
|
|
|
chunking:
|
|
max_size: 2000 # Maximum characters per chunk (2000 = ~50 lines of code)
|
|
# 💡 ADJUST IF: Getting results that are too narrow/broad
|
|
# Small projects: 1500 | Large projects: 3000 | Detailed analysis: 4000
|
|
|
|
min_size: 150 # Minimum characters per chunk (150 = ~4-5 lines)
|
|
# ⚠️ Don't go below 100 or you'll get fragments
|
|
|
|
strategy: semantic # How to split files into chunks
|
|
# 'semantic': Smart splitting (respects functions, classes) - RECOMMENDED
|
|
# 'fixed': Simple splitting (just cuts at size limits) - faster but less intelligent
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 🌊 STREAMING: How we handle really big files
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# Large files (like minified CSS or huge data files) get processed in smaller batches
|
|
# to prevent your computer from running out of memory
|
|
|
|
streaming:
|
|
enabled: true # Always keep this true - prevents memory crashes
|
|
threshold_bytes: 1048576 # Files larger than 1MB use streaming (1MB = 1048576 bytes)
|
|
# 💡 ADJUST IF: Low memory computer = 512000 | High memory = 2097152
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 📁 FILES: Which files to include/exclude from indexing
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
|
|
files:
|
|
min_file_size: 50 # Skip tiny files (50 bytes = ~1 line of code)
|
|
# 💡 REASON: Tiny files usually aren't useful for searching
|
|
|
|
# 🚫 EXCLUDE PATTERNS: Files/folders we always skip (saves time and space)
|
|
exclude_patterns:
|
|
- "node_modules/**" # JavaScript dependencies (huge and not your code)
|
|
- ".git/**" # Git history (not useful for code search)
|
|
- "__pycache__/**" # Python bytecode (generated files)
|
|
- "*.pyc" # More Python bytecode
|
|
- ".venv/**" # Python virtual environments
|
|
- "venv/**" # More virtual environments
|
|
- "build/**" # Compiled output (not source code)
|
|
- "dist/**" # Distribution files
|
|
# 💡 ADD YOUR OWN: Add patterns like "logs/**" or "*.tmp"
|
|
|
|
include_patterns:
|
|
- "**/*" # Include everything else by default
|
|
# 💡 CUSTOMIZE: Could be ["**/*.py", "**/*.js"] for only Python/JS
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 🧠 EMBEDDINGS: How we turn your code into searchable "vectors"
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# Embeddings are like "fingerprints" of your code that help find similar content
|
|
# Don't worry about the technical details - the defaults work great!
|
|
|
|
embedding:
|
|
preferred_method: ollama # Which system to use for creating embeddings
|
|
# 'ollama': Best quality (needs Ollama installed) - RECOMMENDED
|
|
# 'ml': Good quality (downloads models automatically)
|
|
# 'hash': Basic quality (works without internet)
|
|
# 'auto': Try ollama, fall back to ml, then hash - SAFEST CHOICE
|
|
|
|
ollama_model: nomic-embed-text # Which Ollama model to use (this one is excellent)
|
|
ollama_host: localhost:11434 # Where to find Ollama (don't change unless you know why)
|
|
|
|
ml_model: sentence-transformers/all-MiniLM-L6-v2 # Backup model (small and fast)
|
|
|
|
batch_size: 32 # How many chunks to process at once
|
|
# 💡 ADJUST IF: Slow computer = 16 | Fast computer = 64
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 🔍 SEARCH: How the system finds and ranks results
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
|
|
search:
|
|
default_top_k: 10 # How many search results to show by default
|
|
# 💡 MORE RESULTS: 15-20 | FASTER SEARCH: 5-8
|
|
|
|
enable_bm25: true # Also use keyword matching (like Google search)
|
|
# 💡 EFFECT: Finds exact word matches even if semantically different
|
|
# Keep true unless getting too many irrelevant results
|
|
|
|
similarity_threshold: 0.1 # Minimum "similarity score" to show results (0.0-1.0)
|
|
# 💡 HIGHER = fewer but more relevant results
|
|
# Picky: 0.3 | Balanced: 0.1 | Show everything: 0.05
|
|
|
|
expand_queries: false # Automatically add related search terms
|
|
# 💡 EFFECT: "auth" becomes "auth authentication login user"
|
|
# Better results but slower - TUI enables this automatically
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 🤖 LLM: Settings for the AI that explains and synthesizes results
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# The LLM (Large Language Model) reads your search results and explains them in plain English
|
|
|
|
llm:
|
|
ollama_host: localhost:11434 # Where to find Ollama (don't change unless you know why)
|
|
|
|
synthesis_model: auto # Which AI model to use for explanations
|
|
# 'auto': Picks best available model - RECOMMENDED
|
|
# 'qwen3:0.6b': Ultra-fast, good for CPU-only computers
|
|
# 'qwen3:4b': Slower but more detailed explanations
|
|
|
|
expansion_model: auto # Model for query expansion (usually same as synthesis)
|
|
|
|
max_expansion_terms: 8 # How many extra terms to add to expanded queries
|
|
# 💡 MORE TERMS = broader search but potentially less focused
|
|
|
|
enable_synthesis: false # Turn on AI explanations by default
|
|
# 💡 SET TO TRUE: If you want every search to include explanations
|
|
# (You can always use --synthesize flag when you want it)
|
|
|
|
synthesis_temperature: 0.3 # How "creative" the AI explanations are (0.0-1.0)
|
|
# 💡 Lower = more factual | Higher = more creative
|
|
# Code analysis: 0.1-0.3 | Creative writing: 0.7-0.9
|
|
|
|
cpu_optimized: true # Prefer lightweight models for computers without graphics cards
|
|
# 💡 DISABLE IF: You have a powerful GPU and want highest quality
|
|
|
|
enable_thinking: true # Let AI "think out loud" for complex questions
|
|
# 💡 EFFECT: Shows reasoning process, better for learning/debugging
|
|
|
|
#═════════════════════════════════════════════════════════════════════════════════
|
|
# 🎯 QUICK TROUBLESHOOTING:
|
|
#
|
|
# Search returns nothing? → Lower similarity_threshold to 0.05
|
|
# Search too slow? → Set expand_queries: false and batch_size: 16
|
|
# Results not detailed enough? → Increase max_size to 3000
|
|
# Getting weird fragments? → Check min_size is at least 150
|
|
# AI not working? → Make sure Ollama is running: `ollama serve`
|
|
# Out of memory errors? → Decrease batch_size to 16 and lower threshold_bytes
|
|
#═════════════════════════════════════════════════════════════════════════════════ |