g-ntovas/gemma-3-1b-it-gguf-q4_k_m-apostate

Abliterated GGUF Model for MaziyarPanahi/gemma-3-1b-it-GGUF

Quantized GGUF export of MaziyarPanahi/gemma-3-1b-it-GGUF with refusal behavior removed via directional ablation using Apostate.

Details

Parameter	Value
Base model	MaziyarPanahi/gemma-3-1b-it-GGUF
Format	GGUF
Quantization	apostate
File size	768.5 MB

Usage

from apostate import GGUFInference

inference = GGUFInference("g-ntovas/gemma-3-1b-it-gguf-q4_k_m-apostate/gemma-3-1b-it-gguf-q4_k_m-apostate.gguf")
for token in inference.chat_stream([{"role": "user", "content": "Hello!"}]):
    print(token, end="", flush=True)

Citation

@software{apostate,
  title = {Apostate: Inference-Time Refusal Ablation},
  url = {https://github.com/g-ntovas/apostate},
}