license: apache-2.0
tags:
- gemma
- gemma-4
- heretic
- uncensored
- nvfp4
- gguf
- vision
- multimodal
base_model: llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic
Gemma 4 E4B It Ultra Uncensored Heretic - NVFP4 GGUF
NVFP4 quantization of llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic, an abliterated Gemma 4 E4B model with vision support.
About the Model
- Gemma 4 E4B — abliterated/uncensored variant for unrestricted use
- Vision support via separate mmproj vision projector
- Architecture: Gemma 4 with MoE (4B active parameters)
Quantization
Quantized from the BF16 safetensors using llama.cpp (build 537).
NVFP4 (NVIDIA FP4) uses 4-bit floating point quantization optimized for NVIDIA Blackwell GPUs.
Files
| File | Size | Description |
|---|---|---|
gemma-4-E4B-it-ultra-uncensored-heretic-nvfp4.gguf |
~4.8 GB | NVFP4 quantized model |
mmproj-gemma-4-E4B-it-ultra-uncensored-heretic-f16.gguf |
~0.97 GB | Vision projector (BF16) |
Usage
llama-server \
-m gemma-4-E4B-it-ultra-uncensored-heretic-nvfp4.gguf \
--mmproj mmproj-gemma-4-E4B-it-ultra-uncensored-heretic-f16.gguf \
-ngl 99 \
--host 0.0.0.0 \
--port 8080
License
Apache 2.0