Skip to content
Audio

Indic-Transcribe Flex fp16

ASR для английского и индийских языков без лишнего веса

1 min readupdated

Это float16-копия модели bodhan-ai/indic-transcribe-flex от Bodhan AI / AI4Bharat: распознавание речи для английского и индийских языков. В оригинале 1.22B параметров; здесь веса переведены из float32 в float16, остальные файлы побайтно совпадают с оригиналом.

Ключевые особенности

  • Вес model.safetensors уменьшен с 4,89 ГБ до 2,45 ГБ.
  • Конвертация только весов: float32-тензоры стали float16, целочисленные не тронуты, значения проверены на конечность.
  • Публичная копия под той же лицензией: приложению Naavil не нужен вход в Hugging Face для скачивания.
  • В комплекте SHA256-манифест; приложение сверяет каждый файл.
  • Энкодер происходит от nvidia/canary-1b-v2 (CC BY 4.0).

Пример

Сценарий из карточки: офлайн-приложение для диктовки Naavil один раз скачивает копию при настройке, показывает лицензию и проверяет файлы. Проверить скачанный вес можно так:

shasum -a 256 model.safetensors
certutil -hashfile model.safetensors SHA256
ПодробнееПараметров: 1.22B. Размер fp16-весов: 2,45 ГБ; оригинал float32 — 4,89 ГБ.
Лицензия: Indic Open Model License v1.0. Нужна атрибуция «Built with Indic-Transcribe Flex from Bodhan AI / AI4Bharat».
Запуск как API или хостинг-сервис требует письменного разрешения Bodhan AI; для продукта с 500M MAU или $250M годовой выручки — отдельная лицензия.
Бенчмарки и инструкции по запуску — в upstream/README.md.
Discuss Telegram 0 24

Will it run?

Runs in F16

Fits without quantisation.

  • F16weights 2.3 GB · ready files3.0 GB
  • FP8 / INT8recommendedweights 1.1 GB1.8 GB
  • 4-bitweights 0.6 GB1.2 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 2060 · F16 — 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.

    # Rosasynthesiz/indic-transcribe-flex-fp16
    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 Rosasynthesiz/indic-transcribe-flex-fp16 --local-dir ~/models/indic-transcribe-flex-fp16
    # 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. canary-1b-v2family root
  2. indic-transcribe-flex
  3. indic-transcribe-flex-fp16

New posts about versions of canary-1b-v2 (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face

About the model

Rosasynthesiz/indic-transcribe-flex-fp16
Author
Rosasynthesiz
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

240

Ask AI about this model

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

Comments