Instructions to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="fcampanini74/mem-agent-4B-Q4-K-M-GGUF",
	filename="mem-agent-q4_k_m.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Use Docker

docker model run hf.co/fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Ollama:
```
ollama run hf.co/fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
```

Unsloth Studio

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for fcampanini74/mem-agent-4B-Q4-K-M-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for fcampanini74/mem-agent-4B-Q4-K-M-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for fcampanini74/mem-agent-4B-Q4-K-M-GGUF to start chatting

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Docker Model Runner
How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Docker Model Runner:
```
docker model run hf.co/fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M
```

Lemonade

How to use fcampanini74/mem-agent-4B-Q4-K-M-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull fcampanini74/mem-agent-4B-Q4-K-M-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.mem-agent-4B-Q4-K-M-GGUF-Q4_K_M

List all available models

lemonade list

mem-agent 4B - Q4_K_M GGUF

This is a 4-bit GGUF quantization of driaforall/mem-agent, a persistent memory agent trained with online RL.

Model Description

mem-agent is a 4B parameter language model based on Qwen3-4B-Thinking-2507, trained using GSPO (Generalized Supervised Policy Optimization) to interact with a markdown-based memory system inspired by Obsidian.

This GGUF conversion enables:

✅ LM Studio compatibility - Run the model locally with an intuitive GUI
✅ Windows support - Part of a broader mem-agent Windows porting project
✅ CPU/GPU inference - Optimized for various hardware configurations
✅ Reduced memory footprint - ~2GB model size with minimal performance loss

Original Model

The original model was developed by driaforall and achieves impressive results:

75% overall score on md-memory-bench
Rivals models 50x its size on memory tasks
Trained on three core capabilities: Retrieval, Updating, and Clarification

Read the full technical blog post: mem-agent: Persistent, Human Readable Memory Agent

Quantization Details

Format: GGUF (Q4_K_M)
Precision: 4-bit quantization
Size: ~2GB
Performance: Minimal degradation compared to full precision (original 4-bit MLX version: 76.8% overall score)

Why This Port?

The original mem-agent MCP server was designed for Mac and Linux environments. This GGUF conversion is part of a Windows porting project to make mem-agent accessible to a broader audience through:

LM Studio integration for easy local deployment
Cross-platform compatibility
Standard GGUF toolchain support (llama.cpp, Ollama, etc.)

🪟 Windows Port Project

This model is part of the Windows porting effort of mem-agent:

Repository: mem-agent-mcp-Windows
Goal: Enable mem-agent functionality on Windows systems
Integration: Compatible with LM Studio and other GGUF-based tools

Usage

LM Studio

Download the GGUF file
Load it in LM Studio
Configure the model with appropriate system prompts for memory agent functionality

llama.cpp

./main -m mem-agent-4B-Q4-K-M.gguf -p "Your prompt here" -n 512

Downloads last month: 8

GGUF

Model size

4B params

Architecture

qwen3

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for fcampanini74/mem-agent-4B-Q4-K-M-GGUF

Base model

Qwen/Qwen3-4B-Thinking-2507

Quantized

(135)

this model