Skip to content
Computer vision

Huihui-gemma-4-31B-it-abliterated

Gemma 4 31B без отказов

1 min readupdated

Это abliterated-версия google/gemma-4-31B-it на 31,27 млрд параметров. Авторы называют реализацию грубым proof-of-concept: отказы убираются через abliteration без TransformerLens. В этой модели полностью abliterated и thinking, и non-thinking режимы.

Ключевые особенности

  • Удалены отказы в обоих режимах: thinking и non-thinking.
  • Реализация — proof-of-concept abliteration без TransformerLens.
  • Готовый запуск через Ollama: huihui_ai/gemma-4-abliterated:31B.
  • Safety-фильтрация снижена: возможны чувствительные и спорные ответы.
  • Авторы советуют research/тесты и ручную проверку, не production.

Пример

ollama run huihui_ai/gemma-4-abliterated:31B
ПодробнееБазовая модель: google/gemma-4-31B-it, 31,27B параметров.
Бенчмарков в карточке нет.
Требования к VRAM не указаны.
Снижена safety-фильтрация; авторы советуют controlled/research-использование и ручную проверку выводов.
Ollama-тег: huihui_ai/gemma-4-abliterated:31B.
Discuss Telegram 0 15

Will it run?

Runs in 4-bit

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • BF16weights 58 GB · ready files65 GB
  • FP8 / INT8recommendedweights 29 GB33 GB
  • 4-bitweights 16 GB19 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # NobittaRaj/Huihui-gemma-4-31B-it-abliterated
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download NobittaRaj/Huihui-gemma-4-31B-it-abliterated --local-dir ~/models/Huihui-gemma-4-31B-it-abliterated
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

  1. gemma-4-31Bfamily root
  2. gemma-4-31B-it
  3. Huihui-gemma-4-31B-it-abliterated

New posts about versions of gemma-4-31B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
Merges 3All on Hugging Face

About the model

NobittaRaj/Huihui-gemma-4-31B-it-abliterated
Author
NobittaRaj
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

150

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments