Instructions to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="TaQuants/Tema_Q-R-4B-TaQuants-GGUF",
	filename="Tema_Q-R-4B-TaIQ2_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use TaQuants/Tema_Q-R-4B-TaQuants-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 TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
# Run inference directly in the terminal:
llama cli -hf TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
# Run inference directly in the terminal:
llama cli -hf TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_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 TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
# Run inference directly in the terminal:
./llama-cli -hf TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_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 TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M

Use Docker

docker model run hf.co/TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M

LM Studio
Jan

vLLM

How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TaQuants/Tema_Q-R-4B-TaQuants-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TaQuants/Tema_Q-R-4B-TaQuants-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M

Ollama
How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with Ollama:
```
ollama run hf.co/TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
```

Unsloth Studio

How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF 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 TaQuants/Tema_Q-R-4B-TaQuants-GGUF 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 TaQuants/Tema_Q-R-4B-TaQuants-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for TaQuants/Tema_Q-R-4B-TaQuants-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with Docker Model Runner:
```
docker model run hf.co/TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M
```

Lemonade

How to use TaQuants/Tema_Q-R-4B-TaQuants-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull TaQuants/Tema_Q-R-4B-TaQuants-GGUF:IQ2_M

Run and chat with the model

lemonade run user.Tema_Q-R-4B-TaQuants-GGUF-IQ2_M

List all available models

lemonade list

Tema_Q-R-4B TaQuants

The Repository
Technical Report

The Tema_Q development team, team zenei, has developed a new importance matrix method called TaQuants (Tensor-aware Adaptive Quantization).
This model is a TaQuants version of temaq-org/Tema_Q-R-4B created with TaQuants v2.0.

The model size and performance are as follows:
TaIQ2_M is 0.01GB compressed and shows a 0.96% improvement in PPL compared to IQ2_M. TaIQ3_S has a file size increase of 0.16GB compared to IQ3_S. On the other hand, it shows a 3.43% improvement in PPL compared to Q4_K_M, which is 0.35GB larger.