Skip to content
Other

K2-Horizon-7B-Uno heretic

без отказов и подхалимства

1 min read

Поведенческая версия 7B-модели hermitdave/K2-Horizon-7B-Uno-merged — merged LoRA от IFM/K2-Horizon-7B-Uno. Автор репозитория Flakily6416 обработал её в Heretic 2.x методом Arbitrary-Rank Ablation (ARA), чтобы снизить отказы и подхалимство.

Ключевые особенности

  • Снижение отказов: 3/100 на mlabonne/harmful_behaviors test[:100] против 77/100 у исходной модели.
  • Меньше подхалимства: 0/78 на UnstableLlama/praise с льстивым system prompt против до 22/78.
  • Один ARA-проход и 100 триалов Optuna; выбран trial #77 с KL 0.191 относительно источника.
  • Менялись только self_attn.o_proj и mlp.down_proj в learned layer range; layernorms, q_proj и gate_proj бит-в-бит совпадают с источником.
  • Шаблон чата оставлен с reasoning_effort=high; оценка шла с medium и пропуском CoT по тегам <ifm|think*>.

Пример

Сценарий из карточки: на промпт, который обычно вызывает отказ, модель отвечает прямо, без уклончивости; на ложную предпосылку — исправляет её, а не поддакивает. Фактические контрольные ответы не меняются по содержанию и стилю.

ПодробнееОтказы: 3/100 против 77/100; подхалимство: 0/78 против до 22/78; KL к источнику: 0.191.
Обучение: Heretic 2.0.0, ARA, lora_rank=50, 100 триалов Optuna.
Лицензия Apache 2.0.
Примеры и видео
Discuss Telegram 0 87

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 17 GB · ready files19 GB
  • FP8 / INT8recommendedweights 8.4 GB9.7 GB
  • 4-bitweights 4.7 GB5.7 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.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 3080 · BF16 — from RTX 4090

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

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.

    # Flakily6416/K2-Horizon-7B-Uno-heretic
    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 Flakily6416/K2-Horizon-7B-Uno-heretic --local-dir ~/models/K2-Horizon-7B-Uno-heretic
    # 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. K2-Horizon-7Bfamily root
  2. K2-Horizon-7B-Uno
  3. K2-Horizon-7B-Uno-merged
  4. K2-Horizon-7B-Uno-heretic

New posts about versions of K2-Horizon-7B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 2All on Hugging Face

About the model

Flakily6416/K2-Horizon-7B-Uno-heretic
Author
Flakily6416
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

870

Ask AI about this model

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

Comments