Instructions to use bartowski/SmallThinker-3B-Preview-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bartowski/SmallThinker-3B-Preview-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="bartowski/SmallThinker-3B-Preview-GGUF",
	filename="SmallThinker-3B-Preview-IQ2_M.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use bartowski/SmallThinker-3B-Preview-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_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 bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_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 bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M

Use Docker

docker model run hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use bartowski/SmallThinker-3B-Preview-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bartowski/SmallThinker-3B-Preview-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": "bartowski/SmallThinker-3B-Preview-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M

Ollama
How to use bartowski/SmallThinker-3B-Preview-GGUF with Ollama:
```
ollama run hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
```

Unsloth Studio new

How to use bartowski/SmallThinker-3B-Preview-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 bartowski/SmallThinker-3B-Preview-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 bartowski/SmallThinker-3B-Preview-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for bartowski/SmallThinker-3B-Preview-GGUF to start chatting

Docker Model Runner
How to use bartowski/SmallThinker-3B-Preview-GGUF with Docker Model Runner:
```
docker model run hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M
```

Lemonade

How to use bartowski/SmallThinker-3B-Preview-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull bartowski/SmallThinker-3B-Preview-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.SmallThinker-3B-Preview-GGUF-Q4_K_M

List all available models

lemonade list

Q8_0 is as fast as Q2_* in ollama

by aguspiza - opened Dec 30, 2024

Discussion

aguspiza

Dec 30, 2024

•

edited Jan 2, 2025

Q8_0 is as fast as Q2_* in ollama on CPU which does not make any sense
In fact Q8_0 is faster than nearly everything else

bartowski

Owner Jan 2, 2025

that's.. quite interesting ! what are your specs..?

It is possible that the simplicity of Q8_0 is beneficial, can you try Q4_0?

aguspiza

Jan 2, 2025

•

edited Jan 2, 2025

it is a AMD Ryzen 5700U, running ollama version 0.5.4:

ollama run --verbose hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q8_0

hello thinker!

Hello! I'm here to help you with any questions or topics you'd like to discuss. Whether it's about technology, science, history, literature, philosophy, or
anything else, feel free to share your thoughts and we can explore them together.

So, since you started with "hello thinker!", it seems like you're interested in having someone to talk to about various subjects. I'm more than happy to
assist you. Please go ahead and let me know what's on your mind!

total duration: 11.946873s
load duration: 20.6196ms
prompt eval count: 11 token(s)
prompt eval duration: 520ms
prompt eval rate: 21.15 tokens/s
eval count: 99 token(s)
eval duration: 11.405s
eval rate: 8.68 tokens/s

ollama run --verbose hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_0

hello thinker!

Hello there! I'm just a large language model created by Alibaba, and I'm here to help you with any questions or topics you'd like to discuss. Whether it's
about technology, culture, history, science, or anything else, feel free to share your thoughts, and I'll do my best to provide insightful and informative
responses. Let's dive into a topic that interests you!

total duration: 12.2688759s
load duration: 20.6917ms
prompt eval count: 11 token(s)
prompt eval duration: 977ms
prompt eval rate: 11.26 tokens/s
eval count: 79 token(s)
eval duration: 11.27s
eval rate: 7.01 tokens/s

aguspiza

Jan 18, 2025

That seems to be related to ollama 0.5.4. With ollama 0.5.7:

$ ollama run --verbose hf.co/bartowski/SmallThinker-3B-Preview-GGUF:Q4_0

hello thinker!
Hello there, I'm just a helpful assistant here to assist you with any questions or topics you'd like to discuss.
What would you like to know or talk about today?

total duration: 1.901568s
load duration: 21.2136ms
prompt eval count: 11 token(s)
prompt eval duration: 119ms
prompt eval rate: 92.44 tokens/s
eval count: 36 token(s)
eval duration: 1.76s
eval rate: 20.45 tokens/s

/bye
$ ollama --version
ollama version is 0.5.7

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment