Instructions to use rmaapk/nexis4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use rmaapk/nexis4 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 rmaapk/nexis4:Q2_K
# Run inference directly in the terminal:
llama cli -hf rmaapk/nexis4:Q2_K

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf rmaapk/nexis4:Q2_K
# Run inference directly in the terminal:
llama cli -hf rmaapk/nexis4:Q2_K

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf rmaapk/nexis4:Q2_K
# Run inference directly in the terminal:
./build/bin/llama cli -hf rmaapk/nexis4:Q2_K

Use Docker

docker model run hf.co/rmaapk/nexis4:Q2_K

Llama new
LM Studio
Jan
Ollama
How to use rmaapk/nexis4 with Ollama:
```
ollama run hf.co/rmaapk/nexis4:Q2_K
```
Unsloth Desktop

How to use rmaapk/nexis4 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf rmaapk/nexis4:Q2_K

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": "rmaapk/nexis4:Q2_K"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use rmaapk/nexis4 with Docker Model Runner:
```
docker model run hf.co/rmaapk/nexis4:Q2_K
```

Lemonade

How to use rmaapk/nexis4 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull rmaapk/nexis4:Q2_K

Run and chat with the model

lemonade run user.nexis4-Q2_K

List all available models

lemonade list

Hermes Agent

How to use rmaapk/nexis4 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf rmaapk/nexis4:Q2_K

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 rmaapk/nexis4:Q2_K

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use rmaapk/nexis4 with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf rmaapk/nexis4:Q2_K

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 "rmaapk/nexis4:Q2_K" \
  --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"

NEXIS-4 Base Model

This repository holds the text-planning base and original RMAA materials for the NEXIS-4 music project.

Current base model

qwen2.5-0.5b-instruct-q2_k.gguf — Qwen2.5-0.5B-Instruct Q2_K for original lyrics, song sections, and composition planning. It is text-only and does not generate sung audio. Upstream license: Apache-2.0; see LICENSE-APACHE-2.0.txt.

NEXIS-4 voicebank policy

All four NEXIS-4 singer identities (SOLA, VEXA, LUMEN, KAIYA) are to be crafted from synthetic-only training data. No human voice recordings, third-party speaker presets, or pretrained human singer identities are included or used as NEXIS-4 voices. The voicebanks and singing renderer are still to be built.

Target song structure: Intro → Verse 1 → Chorus 1 → Rap Verse → Chorus 2 → Verse 3 → Chorus 3 → Bridge → Outro. Guitar is excluded from the default arrangement.

Licensing

LICENSE (BSL 1.1) covers only RMAA-original NEXIS-4 materials. The Qwen GGUF remains under its upstream Apache-2.0 license. See MODEL_LICENSES.md.

VectorVision data

The vector corpus is stored in vector/parquet.

Official DOI: 10.57967/hf/11502.

Downloads last month: -

GGUF

Model size

0.6B params

Architecture

qwen2

Hardware compatibility

2-bit

Inference Providers NEW

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