Skip to content
Other

Qwen3.8-27B close-steer

вектор, который заставляет модель заканчивать размышления

1 min readupdated

Это экспериментальный аддитивный control vector для Qwen3.8-27B (GGUF-архитектура qwen35, d_model 5120). Он применяется к слоям 28–36 и должен сокращать «пост-решенческое» размышление — циклы вида «Let's maybe… / Let's ensure…» — подталкивая модель к закрытию thinking. Сила 0.5 уже зашита в вектор: использовать при scale 1.0.

Ключевые особенности

  • Вектор для слоёв 28–36; сила 0.5 зашита, рабочий scale — 1.0.
  • Сокращает повторные рассуждения после уже найденного ответа.
  • Подключается к llama-server через --control-vector и --control-vector-layer-range 28 36.
  • Это не веса модели, а steering-вектор; ответы могут стать короче и решительнее, в том числе ошибочно.

Пример

llama-server -m Qwen3.8-27B-UD-Q4_K_XL.gguf \
  --control-vector qwen38-close-a05.gguf \
  --control-vector-layer-range 28 36
Подробнее На 6 held-out runaway-снапшотах по 8 сэмплов alpha 1 закрыл 40/48 за 384 токена против 9/48 без стиринга.
AIME 2025 II-3, один seed: alpha 0.5 дал 82 правильных при 3 120 thinking-токенах; без стиринга — 82 при 11 029.
Малый out-of-sample тест: 20 задач MATH L4–5 — 19/20 в обоих режимах, thinking сокращён на 48%.
Ограничение: экспериментальный вектор; может обрезать рассуждение и ухудшить ответ. Эффект на сложных AIME нестабилен.
Discuss Telegram 0 14

Will it run?

Runs in GGUF

The most precise of the ready quantised files fits.

  • GGUFweights 0.0 GB · ready files0.5 GB

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: GGUF — from RTX 2060

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.

    # mkultrascale/qwen3.8-27b-close-steer
    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 mkultrascale/qwen3.8-27b-close-steer --local-dir ~/models/qwen3.8-27b-close-steer
    # 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

New posts about versions of qwen3.8-27b-close-steer (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

mkultrascale/qwen3.8-27b-close-steer
Author
mkultrascale
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

140

Ask AI about this model

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

Comments