Skip to content
Other

egy-g2p-byt5-accent

IPA для египетского арабского с выбором акцента

2 min read

Экспериментальная G2P-модель для египетского арабского: переводит текст в IPA и позволяет управлять акцентом — Cairo, Sharqia, Sohag или Aswan. Это полностью дообученный google/byt5-small (300M, byte-level) на датасете egy-g2p-dialects.

Ключевые особенности

  • Четыре тега акцента: arz_cairo, arz_sharqia, arz_sohag, arz_aswan.
  • Вход — sentinel-токен акцента перед UTF-8 байтами текста; для кодирования и декодирования нужен g2p_byt5_lib.py из репозитория.
  • На held-out: PER 0.0552, WER 0.2423, точное совпадение предложений 16.08%.
  • Когда эталонные чтения для двух акцентов различаются, модель следует нужному тегу в 98.0% случаев.
  • Слабое место: при одинаковых эталонах модель всё равно выдаёт разные строки в 23.2% пар.

Пример

import sys
from huggingface_hub import snapshot_download
from transformers import T5ForConditionalGeneration
path = snapshot_download('masterofaudio2077/egy-g2p-byt5-accent')
sys.path.insert(0, path)
import g2p_byt5_lib as L
model = T5ForConditionalGeneration.from_pretrained(path)
print(L.g2p(model, ['قال لي إن القهوة جاهزة'], 'arz_sohag', 'cpu', beams=4))
ПодробнееБенчмарки: PER 0.0552, WER 0.2423, exact match 16.08% (beam 4); у oracle «copy Cairo» — 0.0182 / 0.1017 / 61.77%.
Обучение: 1 750 предложений × 4 акцента = 7 000 строк, 12 эпох; опубликованы веса лучшей эпохи (4-я по validation PER).
Ограничения: IPA-разметка сгенерирована LLM и не проверена носителем; в данных только два основных чтения (Cairo = Sharqia, Sohag = Aswan); длинный текст, код-свитчинг, числа, имена и огласованный ввод не тестировались.
Лицензия весов — Apache-2.0.
Discuss Telegram 0 42

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 1.1 GB · ready files1.7 GB
  • BF16weights 0.6 GB1.1 GB
  • FP8 / INT8recommendedweights 0.3 GB0.8 GB
  • 4-bitweights 0.2 GB0.7 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 · F32 — 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.

    # masterofaudio2077/egy-g2p-byt5-accent
    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 masterofaudio2077/egy-g2p-byt5-accent --local-dir ~/models/egy-g2p-byt5-accent
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to egy-g2p-byt5-accent
  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 byt5-small (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 7All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 3All on Hugging Face

About the model

masterofaudio2077/egy-g2p-byt5-accent
Author
masterofaudio2077
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

420

Ask AI about this model

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

Comments

Other

FarmSAM

FarmSAM