Popular GitHub Repositories for Subword Units

A curated list of pretrained sentence and word embedding models

Microsoft.Recognizers.Text provides recognition and resolution of numbers, units, date/time, etc. in multiple languages (ZH, EN, FR, ES, PT, DE, IT, TR, HI, NL. Partial support for JA, KO, AR, SV). Packages available at: https://www.nuget.org/profiles/Recognizers.Text, https://www.npmjs.com/~recognizers.text

Pre-trained subword embeddings in 275 languages, based on Byte-Pair Encoding (BPE)

:zap: TensorFlowASR: Almost State-of-the-art Automatic Speech Recognition in Tensorflow 2. Supported languages that can use characters or subwords

Python library for analysing faces using PyTorch

Soft speech units for voice conversion

HuBERT content encoders for: A Comparison of Discrete and Soft Speech Units for Improved Voice Conversion

Compare SELUs (scaled exponential linear units) with other activations on MNIST, CIFAR10, etc.

Code for "Aligning Linguistic Words and Visual Semantic Units for Image Captioning", ACM MM 2019

An Out-of-the-Box Replication of GANimation using PyTorch, pretrained weights are available!

Builds wordpiece(subword) vocabulary compatible for Google Research's BERT

Code for Mimicking Word Embeddings using Subword RNNs (EMNLP 2017)

? A list of pre-trained BERT models for Japanese with word/subword tokenization + vocabulary construction algorithm information

Official repository for "Speaking Style Conversion With Discrete Self-Supervised Units" (EMNLP 2023). https://arxiv.org/abs/2212.09730

Basic pytorch implementation of NAC/NALU from Neural Arithmetic Logic Units paper by trask et.al

Acoustic models for: A Comparison of Discrete and Soft Speech Units for Improved Voice Conversion

Context-sensitive word embeddings with subwords. In Rust.

This was my Master's project where i was involved using a dataset from Wireless Sensor Data Mining Lab (WISDM) to build a machine learning model to predict basic human activities using a smartphone accelerometer, Using Tensorflow framework, recurrent neural nets and multiple stacks of Long-short-term memory units(LSTM) for building a deep network. After the model was trained, it was saved and exported to an android application and the predictions were made using the model and the interface to speak out the results using text-to-speech API.

Using convolutional neural networks to predict the Coding Units (CUs) depths in HEVC intra-prediction mode, in order to reduce the time of the encoding process in HEVC.

Code for WECHSEL: Effective initialization of subword embeddings for cross-lingual transfer of monolingual language models.