Instructions to use RalphLabsAI/ralph-crowns with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use RalphLabsAI/ralph-crowns 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 RalphLabsAI/ralph-crowns
# Run inference directly in the terminal:
llama cli -hf RalphLabsAI/ralph-crowns

Install from WinGet (Windows)

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

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 RalphLabsAI/ralph-crowns
# Run inference directly in the terminal:
./llama-cli -hf RalphLabsAI/ralph-crowns

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 RalphLabsAI/ralph-crowns
# Run inference directly in the terminal:
./build/bin/llama-cli -hf RalphLabsAI/ralph-crowns

Use Docker

docker model run hf.co/RalphLabsAI/ralph-crowns

LM Studio
Jan
Ollama
How to use RalphLabsAI/ralph-crowns with Ollama:
```
ollama run hf.co/RalphLabsAI/ralph-crowns
```

Unsloth Studio

How to use RalphLabsAI/ralph-crowns 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 RalphLabsAI/ralph-crowns 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 RalphLabsAI/ralph-crowns to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for RalphLabsAI/ralph-crowns to start chatting

How to use RalphLabsAI/ralph-crowns with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf RalphLabsAI/ralph-crowns

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": "RalphLabsAI/ralph-crowns"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use RalphLabsAI/ralph-crowns with Docker Model Runner:
```
docker model run hf.co/RalphLabsAI/ralph-crowns
```

Lemonade

How to use RalphLabsAI/ralph-crowns with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull RalphLabsAI/ralph-crowns

Run and chat with the model

lemonade run user.ralph-crowns-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use RalphLabsAI/ralph-crowns with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf RalphLabsAI/ralph-crowns

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 RalphLabsAI/ralph-crowns

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use RalphLabsAI/ralph-crowns with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf RalphLabsAI/ralph-crowns

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 "RalphLabsAI/ralph-crowns" \
  --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"

Ralph crowns — Qwen3-8B

The reigning crowned compressions from Bittensor netuid 40, one file per bit tier. Every round re-scores the incumbents against a fresh exam; when a crown changes hands, the file here changes with it.

file	tier	bits/weight	size	retention	round	scored artifact
`ralph-qwen3-8b-ternary.gguf`	ternary	1.6095	2.33 GB	0.180233	1	`crazy-m1ner/ralph-qwen3-8b-ternary` @`main`
`ralph-qwen3-8b-sub4.gguf`	sub4	4.0	4.61 GB	0.302385	1	`andreas11112/qwen3-8b-sn40-sub4` @`8c8cfa61be18`

What "retention" is, and what it is not

Retention measures how much of the pinned parent's effect on a third-party observer model each compression reproduces, aggregated over its worst slice of (observer x language x depth) rather than its average. It is a compression-fidelity measure. It is not a capability benchmark, and a high retention does not by itself mean a model is good at anything in particular.

Round record, with the exam, every per-sample measurement and the crown decision: https://huggingface.co/datasets/RalphLabsAI/ralph-v2-rounds/resolve/main/rounds/round-00000001-6ac6aa97163a2707.json

Provenance

Each file is byte-identical to the artifact the round actually scored: it is downloaded from the miner's own repo at the pinned commit named in the signed record, re-hashed, and published only if the hash matches the model_id in that record. crowns.json carries the source repo and revision for every file, so you can fetch the original and check it yourself.

Credit for the weights belongs to the miners named in crowns.json. This repo is a verified mirror with a stable name, not the origin.

Running them

Any llama.cpp-based runner. On iPhone, PocketPal AI and Enclave AI both load GGUF straight from the Hub — search this repo and pick a file by size. Note that an 8B at ~4.6 GB is close to the per-app memory ceiling on iOS and needs a Pro device; the smaller tiers are the ones that fit comfortably.

Downloads last month: 287

GGUF

Model size

8B params

Architecture

qwen3

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Inference Providers NEW

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

Model tree for RalphLabsAI/ralph-crowns

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Quantized

(383)

this model