toastynews

ToastyNews's repositories

Pre-trained ELECTRA from Hong Kong data

Language:PythonApache-2.02600

fastText vectors created from Hong Kong data.

CC-BY-4.02000

Scraped reviews from OpenRice for sentiment analysis. Formatted to use with BERT.

CC-BY-4.0800

Benchmark for Cantonese word segmentation and pos tagging

Language:PythonMIT600

Simple statistical detector for Hong Kongese/Standard Chinese/English languages.

Language:PythonMIT500

Scraped forum threads from LIHKG for categorization task. Formatted to use with BERT.

CC-BY-4.0500

Scraped word definition pairs from words.hk for semantic similarity task. Formatted to use with BERT.

NOASSERTION400

Hongkongese/Cantonese models compatible with CKIP Transformers

Language:PythonGPL-3.0200

Create training files to fine-tune CKIP Transformers

Language:PythonApache-2.0100

Hong Kongese deep learning data set and notebooks forked from bentrevett/pytorch-sentiment-analysis tutorial

Language:Jupyter NotebookMIT100

Facebook's FastText for Java

Language:JavaBSD-3-Clause000

Data for evaluating translation APIs using Hong Kong text.

Language:PythonCC-BY-4.0000

Scraped forum threads from LIHKG for categorization task. Formatted to use with BERT.

CC-BY-4.0000