Releases: DonTizi/rlama
Release v0.1.39
Update to version 0.1.39
Release v0.1.38
flagembeddings version
Release v0.1.37
Update to version 0.1.37
Release v0.1.36
Update to version 0.1.36
Release v0.1.35
Update to version 0.1.35
Release v0.1.34
Release Update Description for v0.1.34
-
Website Monitoring for RAG Systems
- Introduced commands to set up and manage website monitoring:
web-watch: Configure a RAG system to monitor websites for updates.web-watch-off: Disable website monitoring.check-web-watched: Manually check monitored websites for updates.
- Introduced commands to set up and manage website monitoring:
-
Hugging Face Integration
- Access to GGUF models on Hugging Face:
hf-browse: Browse available models.run-hf: Run a Hugging Face GGUF model.
- Examples:
rlama hf-browse mistral --openrlama run-hf bartowski/Llama-3.2-1B-Instruct-GGUF
- Access to GGUF models on Hugging Face:
-
Advanced Chunking Strategies
- New chunking strategies for optimized document retrieval:
fixed,semantic,hybrid,hierarchical
- Examples:
rlama rag llama3 documentation ./docs --chunking-strategy=semanticrlama rag llama3 book-rag ./books --chunking-strategy=hierarchical
- New chunking strategies for optimized document retrieval:
-
OpenAI Model Support
- Use OpenAI models for inference with RLAMA:
- Set API key with
export OPENAI_API_KEY="your-api-key" - Create RAG with OpenAI models:
rlama rag gpt-4-turbo my-rag ./documents
- Supported models include
gpt-4-turbo,o3-mini, etc.
- Set API key with
- Use OpenAI models for inference with RLAMA:
-
Custom Data Directory
- Set a custom data directory for RLAMA:
- Command-line flag:
rlama --data-dir /path/to/directory - Environment variable:
export RLAMA_DATA_DIR=/path/to/directory
- Command-line flag:
- Set a custom data directory for RLAMA:
-
Reranking Enhancements
- Configuration for reranking models:
add-reranker: Configure reranking for a RAG.- Parameters include model, weight, and threshold.
- Example:
rlama add-reranker my-rag --model reranker-model --weight 0.8
- Configuration for reranking models:
-
Profile Management
- Manage multiple API profiles:
profile add: Add a new API profile.profile list: List all profiles.profile delete: Delete a profile.
- Example:
rlama profile add openai-work openai "sk-your-api-key"
- Manage multiple API profiles:
-
Wizard Enhancements
- Improved interactive setup for creating RAG systems:
- Support for website crawling options.
- Enhanced chunking strategy selection.
- Improved interactive setup for creating RAG systems:
-
Documentation Updates
- Added detailed guidelines for chunking strategies and reranking.
-
Various Improvements and Bug Fixes
- Enhanced commands for adding documents.
- Improved error handling and user feedback.
- Updated dependencies and internal libraries for better performance and stability.
Release v0.1.33
fixed Windows uninstall
Release v0.1.32
Update to version 0.1.32
Release v0.1.31
Update to version 0.1.31
Release v0.1.30
RLAMA v0.1.30 - Web Crawling & Interactive Wizard
We're excited to announce RLAMA v0.1.30, bringing powerful new capabilities to your local RAG experience. This release adds web crawling functionality to build RAGs directly from websites and introduces an interactive wizard for easier RAG creation.
🔍 New Web Crawling Features
Build RAG systems directly from websites with our new crawling capabilities:
-
crawl-rag: Create a new RAG system by crawling a websiterlama crawl-rag llama3 docs-rag https://docs.example.com --max-depth=2
-
crawl-add-docs: Add website content to an existing RAGrlama crawl-add-docs my-rag https://blog.example.com --exclude-path=/archive,/tags
Control your crawling with flexible options:
- Set crawling depth with
--max-depth - Adjust concurrent requests with
--concurrency - Skip specific paths with
--exclude-path - Fine-tune chunking with
--chunk-sizeand--chunk-overlap
🧙♂️ Interactive RAG Creation Wizard
Setting up a RAG system is now easier than ever with our step-by-step wizard:
rlama wizardThe wizard guides you through:
- Naming your RAG
- Choosing an Ollama model
- Selecting document folders
- Configuring chunking parameters
- Setting up file filtering
Perfect for new users or anyone who prefers a guided approach to RAG creation!
🔧 Technical Improvements
- Updated Dependencies: Upgraded to Go 1.23.0
- New Libraries: Added support for
goqueryfor HTML parsing and improved document handling - Performance: Enhanced concurrency handling for faster web crawling
Usage Examples
Creating a RAG from a documentation website:
rlama crawl-rag llama3 product-docs https://product.example.com/docs --max-depth=3Adding blog content to an existing knowledge base:
rlama crawl-add-docs knowledge-base https://company.blog.com --concurrency=10Using the interactive wizard:
rlama wizard
# Follow the prompts to create your customized RAGWe're continuously working to make RLAMA the best tool for local RAG systems. As always, your feedback is welcome!