Skip to content
Audio

Praxy LFM3 FR Medical Delib

второй проход для медицинской диктовки

1 min read

Это вторая ступень каскада Praxy для французской медицинской диктовки: модель получает аудио 16 кГц и черновик первого прохода, а затем локально исправляет расшифровку. 1,98 млрд параметров, 24 слоя; веса — 7,4 ГБ в fp32.

Ключевые особенности

  • На входе — аудио и гипотеза первого прохода.
  • Не переписывает фразу целиком, а редактирует черновик.
  • Задержка 33–108 мс на высказывание на H100 в зависимости от нагрузки.
  • Обучена на реальных черновиках: 40 209 диктовок и 19 998 записей общего регистра.
  • Для продакшена рекомендован лексический guardrail.

Пример

prompt = ("Perform ASR in French." +
          "\nFirst-pass hypothesis (to correct): " + brouillon)
conv = [{"role": "system", "content": prompt},
        {"role": "user", "content": [{"type": "audio", "audio": wav}]}]
ПодробнееWER lower_punct на 500 высказываниях: 6,30 для val_drugs_std против 6,47 у первого прохода; общий регистр — 8,42 против 8,95.
Для сравнения: NeMo V6 + T5Gemma2 — 7,41 и 6,76; второй проход быстрее T5Gemma2 (p50 111 мс, p95 194–213 мс).
Ограничения: на уже идеальном черновике возможна деградация до 1,6 пункта; на очень плохих черновиках (29,8% WER) выигрыш всего 2,9 пункта. Нужен лексический guardrail.
Лицензия LFM3 (Liquid AI). Веса 7,4 ГБ в fp32; замеры на H100.
Discuss Telegram 0 34

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.

    # PraxySante/Praxy-LFM3-FR-Medical-Delib
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    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 PraxySante/Praxy-LFM3-FR-Medical-Delib --local-dir ~/models/Praxy-LFM3-FR-Medical-Delib
    # 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 Praxy-LFM3-FR-Medical-Delib (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

PraxySante/Praxy-LFM3-FR-Medical-Delib
Author
PraxySante
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

340

Ask AI about this model

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

Comments

Language models

MedReason

MedReason