title	ParamMem Reasoning Dashboard
emoji	🧠
colorFrom	blue
colorTo	purple
sdk	docker
app_port	7860

🧠 ParamMem: Augmenting Language Agents with Parametric Reflective Memory

ParamMem is a research-grade architecture designed to overcome reflection collapse in LLM agents. By replacing traditional retrieval-heavy memory with a lightweight parametric encoding module, the agent internalizes cross-sample reasoning patterns to generate diverse and effective corrective signals.

🚀 Key Features

• Neural Trace Visualization: A real-time, "cyber-dark" dashboard built with React and Framer Motion.
• Parametric Reflection: LoRA-tuned memory module that provides high-diversity feedback without retrieval overhead.
• Hybrid Architecture: Combines Episodic, Parametric, and Cross-Sample (ChromaDB) memory systems.
• Lightning Fast: Optimized for the Groq API, supporting models like GPT-OSS and Llama-3.

🛠️ Architecture

• Agent: ParamAgent manages the iterative reasoning loop.
• Memory: ParametricMemory (Local LoRA or Remote API) handles the internal "logic of reflection."
• Retrieval: CrossSampleMemory utilizes ChromaDB for pattern retrieval.
• Backend: FastAPI server providing low-latency solve endpoints.
• Frontend: Modern React dashboard with interactive iteration depth and memory toggles.

📦 Installation & Setup

Prerequisites

• Python 3.10+
• Node.js 20+
• Groq API Key (for high-speed mode)

Local Development

1. Clone the repository:

   git clone https://github.com/YOUR_USERNAME/ParamMem.git
   cd ParamMem

2. Backend Setup:

   pip install -r requirements.txt
   cp .env.example .env
   # Edit .env with your GROQ_API_KEY
   python -m uvicorn src.api.server:app --reload

3. Frontend Setup:

   cd src/web
   npm install
   npm run dev

🐳 Deployment (Hugging Face Spaces)

The project is container-ready. Use the provided Dockerfile or run the automated script:

python deploy_to_hf.py

📖 Research & Citation

This project is part of ongoing research into reflective language agents.

Title: ParamMem: Augmenting Language Agents with Parametric Reflective Memory
Conference: ICLR 2025 (Projected)

@article{parammem2025,
  title={ParamMem: Augmenting Language Agents with Parametric Reflective Memory},
  author={ParamMem Research Group},
  year={2025}
}

📜 License

Distributed under the MIT License. See LICENSE for more information.