Instructions to use EREN121232/THUNDER-AI-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use EREN121232/THUNDER-AI-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="EREN121232/THUNDER-AI-GGUF",
	filename="THUNDER-AI-R1 V1.2 1.5B.Q4_K_M.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 EREN121232/THUNDER-AI-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf EREN121232/THUNDER-AI-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf EREN121232/THUNDER-AI-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 EREN121232/THUNDER-AI-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf EREN121232/THUNDER-AI-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 EREN121232/THUNDER-AI-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf EREN121232/THUNDER-AI-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 EREN121232/THUNDER-AI-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf EREN121232/THUNDER-AI-GGUF:Q4_K_M

Use Docker

docker model run hf.co/EREN121232/THUNDER-AI-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use EREN121232/THUNDER-AI-GGUF with Ollama:
```
ollama run hf.co/EREN121232/THUNDER-AI-GGUF:Q4_K_M
```

Unsloth Studio new

How to use EREN121232/THUNDER-AI-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 EREN121232/THUNDER-AI-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 EREN121232/THUNDER-AI-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for EREN121232/THUNDER-AI-GGUF to start chatting

Docker Model Runner
How to use EREN121232/THUNDER-AI-GGUF with Docker Model Runner:
```
docker model run hf.co/EREN121232/THUNDER-AI-GGUF:Q4_K_M
```

Lemonade

How to use EREN121232/THUNDER-AI-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull EREN121232/THUNDER-AI-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.THUNDER-AI-GGUF-Q4_K_M

List all available models

lemonade list

THUNDER-AI-GGUF

THUNDER-AI-GGUF is a GGUF release of the THUNDER AI model for local inference.

Available model file

THUNDER-AI-R1 V1.2 1.5B.Q4_K_M.gguf

Ollama usage

Recommended: build and run the cleaned Ollama wrapper:

ollama create thunder-ai-clean -f Modelfile.thunder-clean
ollama run thunder-ai-clean

If you run the raw GGUF directly from Hugging Face:

ollama run hf.co/EREN121232/THUNDER-AI-GGUF:Q4_K_M

you may see visible reasoning blocks like <think>...</think>, because that is the raw reasoning model output path.

Included helper files

Modelfile.thunder-clean
- Builds a cleaned Ollama wrapper model that avoids leaking <think>...</think> tags.
- Also sets num_ctx 8192 for a larger working context.
ollama_memory_proxy.py
- Optional local proxy for Ollama-compatible clients.
- Adds lightweight conversation memory by saving useful facts/preferences from user messages and injecting relevant memories into future prompts.

Build the cleaned Ollama model

ollama create thunder-ai-clean -f Modelfile.thunder-clean

This is the recommended Ollama entrypoint for normal chat usage.

Optional memory proxy usage

The memory proxy is meant for local setups where an app talks to Ollama through an HTTP endpoint.

Set these environment variables if you want to customize it:

THUNDER_REAL_OLLAMA_BASE_URL
THUNDER_PROXY_HOST
THUNDER_PROXY_PORT
THUNDER_MEMORY_FILE
THUNDER_MEMORY_MAX
THUNDER_MEMORY_INJECT_MAX

Then run:

python ollama_memory_proxy.py

By default it listens on 127.0.0.1:11435 and forwards requests to Ollama on 127.0.0.1:11434.

Notes

This repo is for local GGUF usage.
Machine-specific launcher scripts were intentionally not included in the repo because they depend on local Windows paths and drive layout.
The model was fine-tuned and exported with Unsloth.

Downloads last month: 40

GGUF

Model size

2B params

Architecture

qwen2

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 1 Ask for provider support