https://github.com/mrzarei5/modelboard
Conversational agentic RAG app to discover, search, and compare open-source LLMs using local leaderboard data.
https://github.com/mrzarei5/modelboard
agentic-ai agentic-rag huggingface langchain leaderboard llm openai rag semantic-search
Last synced: about 1 month ago
JSON representation
Conversational agentic RAG app to discover, search, and compare open-source LLMs using local leaderboard data.
- Host: GitHub
- URL: https://github.com/mrzarei5/modelboard
- Owner: mrzarei5
- License: apache-2.0
- Created: 2025-06-15T07:28:42.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2025-06-15T07:49:53.000Z (12 months ago)
- Last Synced: 2025-06-15T08:37:57.913Z (12 months ago)
- Topics: agentic-ai, agentic-rag, huggingface, langchain, leaderboard, llm, openai, rag, semantic-search
- Language: Python
- Homepage:
- Size: 3.44 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ModelBoard: Agentic RAG for LLM Leaderboard Discovery
A modular conversational app for searching, filtering, and comparing top open-source large language models, powered by agentic RAG (retrieval-augmented generation) and a curated leaderboard database.
---
*Interactive semantic search and model comparison, powered by agentic RAG.*
---
## 🚀 Features
* **Conversational agentic interface** for model discovery, search, and comparison.
* **Semantic search** across curated leaderboard models (Open LLM Leaderboard, Hugging Face).
* **Advanced filtering** by task, tags, provider, license, benchmark score, parameter count, etc.
* **Model comparison** with detailed metadata, benchmark, and license info.
* **Tool-using agent** architecture (LangChain) with modular, extensible codebase.
* **Local, fast, and reproducible**: all model data indexed and embedded locally—no need to hit Hugging Face API at query time.
* **Streamlit web UI**—easy to run, easy to extend.
---
## 🛠️ Getting Started
### Option 1: Local Python/Conda Installation
1. **Clone and install**
```bash
git clone https://github.com/mrzarei5/ModelBoard.git
cd modelboard-agentic-rag
# Create and activate a conda environment
conda create -n modelboard python=3.10
conda activate modelboard
# Install requirements
pip install -r requirements.txt
```
2. **Set your API keys**
Create a `.env` file in the project root directory and add your OpenAI API key:
```ini
OPENAI_API_KEY=sk-...
```
(Optional) Add your Hugging Face API key:
```ini
HF_API_KEY=hf-...
```
3. **(Optional) Update model leaderboard metadata**
To download or refresh the leaderboard data:
```bash
python data/fetch_leaderboard.py
```
This will (re)generate `model_metadata.json` in the `data` folder.
4. **Run the app**
```bash
streamlit run main.py
```
---
### Option 2: Run with Docker
1. **Clone the repository**
```bash
git clone https://github.com/mrzarei5/ModelBoard.git
cd ModelBoard
```
2. **Create a `.env` file**
In the project root, create a file named `.env` with your API keys:
```ini
OPENAI_API_KEY=sk-...
# (Optional) Hugging Face API key:
# HF_API_KEY=hf-...
```
3. **Build the Docker image**
```bash
docker build -t modelboard .
```
4. **(Optional) Update model leaderboard metadata inside the container**
```bash
docker run --rm -v $(pwd)/data:/app/data modelboard python data/fetch_leaderboard.py
```
5. **Run the app**
```bash
docker run -p 8501:8501 -v $(pwd)/data:/app/data modelboard
```
- The app will be available at: [http://localhost:8501](http://localhost:8501)
- For persistent leaderboard data, mount the `data/` directory as shown above.
---
## 🗂️ Example Queries
* *Show all chat models with Apache-2.0 license.*
* *Find a multilingual model for question answering.*
* *Compare meta-llama/Llama-2-70b-chat-hf and mistralai/Mistral-7B-Instruct-v0.3.*
* *Give me details about openchat/openchat-3.5-0106.*
---
## 📝 License
This project is licensed under the Apache License, Version 2.0.
See the [LICENSE](LICENSE) file for details.
---
## 🤝 Acknowledgements
* [Hugging Face Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
* [LangChain](https://github.com/langchain-ai/langchain)
* [ChromaDB](https://www.trychroma.com/)
* [SentenceTransformers](https://www.sbert.net/)
* [Streamlit](https://streamlit.io/)