Instructions to use HeavensHackDev/Gear-1-160m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HeavensHackDev/Gear-1-160m with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="HeavensHackDev/Gear-1-160m",
	filename="gear-1-160m.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use HeavensHackDev/Gear-1-160m with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf HeavensHackDev/Gear-1-160m
# Run inference directly in the terminal:
llama-cli -hf HeavensHackDev/Gear-1-160m

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf HeavensHackDev/Gear-1-160m
# Run inference directly in the terminal:
llama-cli -hf HeavensHackDev/Gear-1-160m

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 HeavensHackDev/Gear-1-160m
# Run inference directly in the terminal:
./llama-cli -hf HeavensHackDev/Gear-1-160m

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 HeavensHackDev/Gear-1-160m
# Run inference directly in the terminal:
./build/bin/llama-cli -hf HeavensHackDev/Gear-1-160m

Use Docker

docker model run hf.co/HeavensHackDev/Gear-1-160m

LM Studio
Jan

vLLM

How to use HeavensHackDev/Gear-1-160m with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HeavensHackDev/Gear-1-160m"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HeavensHackDev/Gear-1-160m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/HeavensHackDev/Gear-1-160m

Ollama
How to use HeavensHackDev/Gear-1-160m with Ollama:
```
ollama run hf.co/HeavensHackDev/Gear-1-160m
```

Unsloth Studio new

How to use HeavensHackDev/Gear-1-160m 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 HeavensHackDev/Gear-1-160m 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 HeavensHackDev/Gear-1-160m to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for HeavensHackDev/Gear-1-160m to start chatting

Docker Model Runner
How to use HeavensHackDev/Gear-1-160m with Docker Model Runner:
```
docker model run hf.co/HeavensHackDev/Gear-1-160m
```

Lemonade

How to use HeavensHackDev/Gear-1-160m with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull HeavensHackDev/Gear-1-160m

Run and chat with the model

lemonade run user.Gear-1-160m-{{QUANT_TAG}}

List all available models

lemonade list

Model Card for Gear-1-160M

Model Details

Model Description

Gear-1-160M is a small Transformer LLM with about 160 million parameters in GGUF format, designed to run fast on local machines with low memory (CPU/GPU). It’s good for simple chat and basic tasks, but it may be slow or make mistakes since this is my first attempt at making a neural network. I plan to improve it in the future, so please bear with it. A 300M-parameter version will be coming soon.

Developed by: HeavensHack
Model type: GPTNeoX (general.architecture = gptneox).
Language(s) (NLP): English
License: MIT

Uses

Direct Use

Simple question answering

Out-of-Scope Use

Complex reasoning tasks
High-stakes decision making
Content requiring very high accuracy

Bias, Risks, and Limitations

May make mistakes or “hallucinate” answers
First experimental version, so behavior can be inconsistent
Not suitable for critical tasks

Recommendations

Use for learning, demos, or hobby projects
Do not rely on it for serious or sensitive tasks

How to Get Started

Load the GGUF model in your preferred LLM inference framework (LMStudio)
Designed for CPU/GPU local inference

Training Details

Training Data: Custom dataset / small-scale experimental
Training Procedure: First experimental training, not fully optimized
Training Regime: Mixed precision (FP16/FP32)

Evaluation

Tested on small chat prompts
Can respond to simple queries
Results may vary; not benchmarked for production use

Environmental Impact

Small-scale training, low compute and low carbon footprint

Model Card Contact

Author: HeavensHackDev

But...

At first, only the GGUF file will be available. The rest will follow later.

Downloads last month: 3

GGUF

Model size

0.2B params

Architecture

gptneox

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Collection including HeavensHackDev/Gear-1-160m

Gear

Collection

The Gear family of LLMs • 2 items • Updated Dec 6, 2025