Instructions to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M

Use Docker

docker model run hf.co/baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with Ollama:
```
ollama run hf.co/baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
```
Unsloth Desktop

How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with Docker Model Runner:
```
docker model run hf.co/baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M
```

Lemonade

How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Nemotron-3-Super-120B-A12B-MINT-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use baa-ai/Nemotron-3-Super-120B-A12B-MINT-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf baa-ai/Nemotron-3-Super-120B-A12B-MINT-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 "baa-ai/Nemotron-3-Super-120B-A12B-MINT-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"

tomkay commited on Mar 27

Commit

3bbba30

verified ·

1 Parent(s): 4e9e63c

Add model card

Browse files

Files changed (1) hide show

README.md +59 -0

README.md ADDED Viewed

	@@ -0,0 +1,59 @@

+---
+library_name: gguf
+tags:
+  - gguf
+  - quantized
+  - mint
+license: other
+license_name: nvidia-open-model-license
+base_model: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
+base_model_relation: quantized
+---
+<p align="center">
+  <img src="https://huggingface.co/spaces/baa-ai/MINT/resolve/main/baa-logo.svg" width="300" alt="baa.ai">
+</p>
+# Nemotron-3-Super-120B-A12B-MINT-GGUF
+GGUF quantized version of [nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16) using [MINT](https://github.com/baa-ai/MINT).
+> Hybrid Mamba-MoE-Attention architecture (512 experts, 22 active per token) — Q4_K_M quantized for llama.cpp.
+## Files
+| File | Quant | Size |
+|------|-------|------|
+| `Nemotron-3-Super-120B-Q4_K_M.gguf` | Q4_K_M | 80 GB |
+## Metrics
+| Metric | Value |
+|--------|-------|
+| **Size** | **80 GB** |
+| Quantization | Q4_K_M |
+| Framework | llama.cpp (GGUF) |
+| Architecture | Hybrid Mamba-2 + MoE + Attention |
+| Parameters | 123.6B (12B active) |
+## Usage
+```bash
+llama-cli -m Nemotron-3-Super-120B-Q4_K_M.gguf \
+    -p "Hello!" -n 256 --threads 8
+```
+## Notes
+- Requires **llama.cpp build ≥ 8500** for NemotronH MoE + latent projection support
+- Peak memory: ~85 GB
+- MTP (Multi-Token Prediction) layers are stripped during conversion
+## About MINT
+MINT (Memory-Informed N-bit Tuning) formulates mixed-precision quantization as a budget-constrained optimization problem over per-tensor rate-distortion curves. The entire pipeline is data-free.
+- [Paper](https://baa.ai/articles/24-mint-paper.html) | [Code](https://github.com/baa-ai/MINT) | [Models](https://huggingface.co/baa-ai)
+---
+*Quantized by [baa.ai](https://baa.ai)*