fatih-can/Kumru-2B-Abliterated-GGUF

Kumru-2B-Abliterated (GGUF)

A decensored (abliterated) variant of **Kumru-2B**, a 2B-parameter Turkish instruction-tuned language model.

This repo contains GGUF quantizations of the abliterated model for use with llama.cpp and its many front-ends (Ollama, LM Studio, llama-cpp-python, etc.). The full-precision merged weights are in the sibling repo `fatih-can/Kumru-2B-Abliterated`. If you want the original HuggingFace safetensors format, use that repo instead.

Quantized files

Usage with llama.cpp

# Q4_K_M (recommended default)
llama-cli -m Kumru-2B-Abliterated-Q4_K_M.gguf \
  -p "Merhaba, kendinden kısaca bahseder misin?" \
  --temp 0.7 --top-p 0.9

# Chat / server API (Ollama-style OpenAI endpoint)
llama-server -m Kumru-2B-Abliterated-Q4_K_M.gguf \
  --host 0.0.0.0 --port 8080 --jinja

Or import into Ollama:

ollama create kumru-2b-abliterated -f Modelfile
# Modelfile:
FROM ./Kumru-2B-Abliterated-Q4_K_M.gguf

Usage warnings

• —Sensitive or controversial outputs: safety filtering has been significantly reduced; the model may produce sensitive, controversial, or inappropriate content. Review outputs carefully.
• —Not suitable for all audiences / public or commercial production use without additional safeguards.
• —Legal and ethical responsibility: ensure your usage complies with applicable laws and ethical standards; you are solely responsible for any consequences.
• —This is a decensored research artifact and has not undergone safety optimization.

License

Apache-2.0 (matching the base model).