metadata

license: mit
language:
  - ru
library_name: transformers
tags:
  - text-generation-inference

text-normalization-ru-new

Normalization for Russian text. Couldn't find any existing solutions (besides algorithms, don't like those) so made this. It was designed for Silero TTS model which cant handle english and numbers for russian text to speach.

This model is a fine-tuned version of cointegrated/rut5-small on https://www.kaggle.com/c/text-normalization-challenge-russian-language and additional dataset prepared by me using typical messages.

It achieves the following results on the evaluation set:

Loss: 0.0177
Mean Distance: 0
Max Distance: 15

Model description

Tiny T5 trained from scratch for normalizing Russian texts:

translating numbers into words
expanding abbreviations into phonetic letter combinations
transliterating english into russian letters
whatever else was in the dataset (see below)