---

license: apache-2.0

base_model: densenet/Gemma-4-31B-StyleTune-heretic-ara

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

language:

- en

tags:

- gguf

- quantized

- gemma4

---

Gemma-4-31B StyleTune heretic-ara i1-Q3_K_S

This repository publishes the local file Gemma-4-31B-StyleTune-heretic-ara.i1-Q3_K_S.gguf (14,366,910,080 bytes). groxaxo did not quantize it.

The GGUF header records:

• general.architecture: gemma4
• general.name: Gemma 4 31B StyleTune Heretic Ara
• general.finetune: StyleTune-heretic-ara
• general.license: apache-2.0
• general.file_type: 11
• general.url: https://huggingface.co/mradermacher/Gemma-4-31B-StyleTune-heretic-ara-i1-GGUF
• general.source.url: https://huggingface.co/densenet/Gemma-4-31B-StyleTune-heretic-ara
• general.base_model.0.repo_url: https://huggingface.co/Gryphe/Gemma-4-31B-StyleTune
• quantize.imatrix.file: Gemma-4-31B-StyleTune-heretic-ara-i1-GGUF/Gemma-4-31B-StyleTune-heretic-ara.imatrix.gguf
• quantize.imatrix.dataset: imatrix-training-full-3
• quantize.imatrix.entries_count: 410
• quantize.imatrix.chunks_count: 320

The upstream quant repository and the source model card both report Apache-2.0. This file is the i1-Q3_K_S quant from that mradermacher repository. It is not the separate groxaxo/Gemma-4-31B-StyleTune-UD-Q4_K_XL quant. SHA256SUMS is the SHA-256 of the uploaded file.

Use

Load the GGUF with llama.cpp or another GGUF runner. No new quality measurement was made for this upload.