Instructions to use reedmayhew/Grok-3-gemma3-4B-distilled with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use reedmayhew/Grok-3-gemma3-4B-distilled with Transformers:

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("reedmayhew/Grok-3-gemma3-4B-distilled", dtype="auto")

llama-cpp-python

How to use reedmayhew/Grok-3-gemma3-4B-distilled with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="reedmayhew/Grok-3-gemma3-4B-distilled",
	filename="gemma-3-finetune.Q8_0.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use reedmayhew/Grok-3-gemma3-4B-distilled with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
# Run inference directly in the terminal:
llama-cli -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
# Run inference directly in the terminal:
llama-cli -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

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 reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

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 reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

Use Docker

docker model run hf.co/reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

LM Studio
Jan
Ollama
How to use reedmayhew/Grok-3-gemma3-4B-distilled with Ollama:
```
ollama run hf.co/reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
```

Unsloth Studio new

How to use reedmayhew/Grok-3-gemma3-4B-distilled 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 reedmayhew/Grok-3-gemma3-4B-distilled 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 reedmayhew/Grok-3-gemma3-4B-distilled to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for reedmayhew/Grok-3-gemma3-4B-distilled to start chatting

Docker Model Runner
How to use reedmayhew/Grok-3-gemma3-4B-distilled with Docker Model Runner:
```
docker model run hf.co/reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0
```

Lemonade

How to use reedmayhew/Grok-3-gemma3-4B-distilled with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull reedmayhew/Grok-3-gemma3-4B-distilled:Q8_0

Run and chat with the model

lemonade run user.Grok-3-gemma3-4B-distilled-Q8_0

List all available models

lemonade list

xAI Grok 3

Distilled - Gemma 3 4B

NEW REASONING VERSION AVAILABLE:

https://huggingface.co/reedmayhew/Grok-3-reasoning-gemma3-4B-distilled-GGUF

Overview

This model is a Gemma 3 4B variant distilled from xAI’s Grok 3. It was fine-tuned to emulate Grok’s depth and structured clarity, particularly in tasks involving complex thought, such as problem-solving, coding, and mathematics.

Technical Details

Developed by: reedmayhew
Base Model: google/gemma-3-4b-it
Training Speed Enhancement: Trained 2x faster with Unsloth and Huggingface's TRL library

Training Data

The model was trained on:

reedmayhew/Grok-3-100x

This dataset consists of 100 high-quality Grok 3 completions responding to deep questions, solving math problems, and writing or analyzing code. The aim was to distill Grok’s analytical approach and technical versatility into a smaller, accessible model.

This Gemma model was trained 2x faster with Unsloth and Huggingface's TRL library.

Downloads last month: 256

GGUF

Model size

4B params

Architecture

gemma3

Hardware compatibility

8-bit

Inference Providers NEW

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

Model tree for reedmayhew/Grok-3-gemma3-4B-distilled

Base model

google/gemma-3-4b-pt

Finetuned

google/gemma-3-4b-it

Finetuned

unsloth/gemma-3-4b-it

Quantized

(17)

this model

reedmayhew
/

Grok-3-gemma3-4B-distilled