Skip to content
Computer vision

VEHME-FERMAT

проверка рукописных решений по математике

1 min read

VEHME-FERMAT — vision-language модель от исследователей UNIST и POSTECH для оценки открытых рукописных решений по математике. На вход подаются изображение задачи, эталонный ответ и рукописное решение ученика; модель определяет, есть ли ошибка, и локализует её. Построена на Qwen2.5-VL-7B-Instruct и дообучена LoRA на подмножестве FERMAT.

Ключевые особенности

  • Работает с грязной рукописью и многошаговыми решениями, без предварительной расшифровки в текст.
  • Две задачи: Error Detection (ED) и Error Localization (EL).
  • При ошибке выдаёт рассуждение, метку корректности и указание на конкретное выражение или шаг.
  • Пайплайн: SFT на distilled reasoning traces от QwQ-32B + RL GRPO; EVPM на YOLO для spatial grounding.
  • Ответ в формате <think>, <correctness> и <localization>; язык карточки — английский.

Пример

from transformers import AutoProcessor, Qwen2_5_VLForConditionalGeneration
MODEL_ID = "phuongnt/VEHME-FERMAT"
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype="auto", device_map="auto")
processor = AutoProcessor.from_pretrained(MODEL_ID)
# далее — изображение задачи/решения и промпт с эталонным ответом
ПодробнееAIHub: ED Acc 73.01%, EL Acc 61.13%; FERMAT: ED Acc 62.61%, EL Acc 31.90%.
База — Qwen2.5-VL-7B-Instruct; LoRA на MLP-слоях (rank 8, alpha 32), около 6 NVIDIA A100 GPU-days.
На FERMAT проприетарные модели всё ещё сильнее; данные AIHub не перераспространяются.
Лицензия Apache 2.0; FERMAT — CC BY 4.0.
Discuss Telegram 0 22

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.

    # phuongnt/VEHME-FERMAT
    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 phuongnt/VEHME-FERMAT --local-dir ~/models/VEHME-FERMAT
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to VEHME-FERMAT
  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 VEHME-FERMAT (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

phuongnt/VEHME-FERMAT
Author
phuongnt
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

220

Ask AI about this model

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

Comments