Skip to content
Computer vision

HoloCT-8B

понимание и сегментация 3D-КТ в одной модели

2 min read

HoloCT-8B от Alibaba DAMO Academy — старшая модель семейства HoloCT на языковом бэкбоне Qwen3-VL-8B (9,04 млрд параметров). На вход принимает 3D-объём КТ в NIfTI и текстовый промпт на английском, а на выходе даёт текст и, для grounding-запросов, 3D-бинарные маски. Задачи: скрининг поражений, VQA, генерация отчётов и сегментация.

Ключевые особенности

  • Единая интерпретация КТ: скрининг поражений, VQA и генерация отчётов по текстовому промпту.
  • Grounded-генерация: текстовые находки связываются с воксельными 3D-масками.
  • Сегментация по свободному тексту: анатомические цели и находки, описанные в промпте.
  • Архитектура: глобальный энкодер на базе Med3DVLM, детальный энкодер на базе VISTA3D, языковой бэкбон и 3D mask decoder.
  • Парный инференс: в демо можно сравнить grounded- и ungrounded-ответы на одном КТ.

Пример

CUDA_VISIBLE_DEVICES=0 python -m inference.scripts.demo_vqa \
  --model weights/HoloCT-8B \
  --ct demo_ct/my_ct.nii.gz \
  --prompt "Is there a liver lesion?" \
  --grounded-prompt "Is there a liver lesion? Ground your answer with the mask." \
  --output outputs/holoct_8b_vqa
ПодробнееБенчмарки: в карточке приведены только примеры метрик — Dice, HD95, качество текста и grounded-отчётов — без итоговых цифр.
Железо: Ubuntu 22.04, Python 3.10.16, NVIDIA GPU с bfloat16; демо идут на одной GPU через PyTorch SDPA. VRAM не указан.
Ограничения: промпты на английском, вход — 3D-КТ в NIfTI в единицах Хаунсфилда.
Генерация: до 1024 новых токенов, seed 42; для отчётов temperature 0.15, top-p 0.85. Лицензия MIT.
Discuss Telegram 2 6

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

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.

    # Alibaba-DAMO-Academy/HoloCT-8B
    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 Alibaba-DAMO-Academy/HoloCT-8B --local-dir ~/models/HoloCT-8B
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to HoloCT-8B
  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 HoloCT-8B (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

Alibaba-DAMO-Academy/HoloCT-8B
Author
Alibaba-DAMO-Academy
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

62

Ask AI about this model

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

Comments

Computer vision

PANORAMA-2B

PANORAMA-2B
Computer vision

Mage Hand

Mage Hand
Computer vision

Viper-9B

Viper-9B