Malaya

1. Improved Regex for urls.
2. Now `predict_words` able to do in Jupyter Notebook.

1. Improved sentiment module, now default label is `['negative', 'neutral', 'positive']`, and use better dataset iterate using active learning, https://malaya.readthedocs.io/en/latest/load-sentiment.html
2. Dataset can get at https://github.com/huseinzol05/malay-dataset/tree/master/sentiment/semisupervised-twitter-3class, label studio labelling for general tweets at https://label.malaysiaai.ml/projects/12/data, label studio labelling for political tweets at https://label.malaysiaai.ml/projects/16/data, get access at https://github.com/malaysia-ai/label-studio#how-to-get-access

1. Added T5 Tatabahasa, https://malaya.readthedocs.io/en/latest/load-tatabahasa.html
2. Added 5MB T5 for True Case, https://malaya.readthedocs.io/en/latest/load-tatabahasa.html
3. Added 5MB T5 for Segmentation, https://malaya.readthedocs.io/en/latest/load-segmentation.html
4. Added FastFormer BASE and TINY for Emotion Analysis, https://malaya.readthedocs.io/en/latest/load-emotion.html
5. Added FastFormer BASE and TINY for Relevancy Analysis, https://malaya.readthedocs.io/en/latest/load-relevancy.html, can infer up to 2048 tokens.
6. Added FastFormer BASE and TINY for Sentiment Analysis, https://malaya.readthedocs.io/en/latest/load-sentiment.html
7. Added FastFormer BASE and TINY for Subjectivity Analysis, https://malaya.readthedocs.io/en/latest/load-subjectivity.html
8. Added FastFormer BASE and TINY for Toxicity Analysis, https://malaya.readthedocs.io/en/latest/load-toxic.html
9. Added FastFormer BASE and TINY for Entity Tagging, https://malaya.readthedocs.io/en/latest/load-entities.html
10. Added Pretrained FastFormer, https://github.com/huseinzol05/malaya/tree/master/pretrained-model/fastformer

1. Added NeuSpell Spelling Correction using T5-Bahasa, https://malaya.readthedocs.io/en/latest/load-spell-correction.html#List-available-Transformer-models
2. Added JamSpell Spelling Correction interface, https://malaya.readthedocs.io/en/latest/load-spell-correction.html#Load-JamSpell-speller
3. Added LibreOffice pEJAm Spelling Correction interface, https://malaya.readthedocs.io/en/latest/load-spell-correction.html#Load-Spylls-speller
4. Added T5 models for segmentation, https://malaya.readthedocs.io/en/latest/load-segmentation.html#Load-Transformer-model
5. Added T5 models for True Case, https://malaya.readthedocs.io/en/latest/load-true-case.html#List-available-Transformer-model
6. Added quantized models for GPT2, https://malaya.readthedocs.io/en/latest/load-prefix-generator.html#Load-GPT2

1. Added NeuSpell based using T5-Bahasa, https://malaya.readthedocs.io/en/latest/load-spell-correction.html#List-available-Transformer-models
2. Added T5 models for segmentation, https://malaya.readthedocs.io/en/latest/load-segmentation.html#Load-Transformer-model

1. Improved Abstractive Summarization dataset and module, https://malaya.readthedocs.io/en/latest/load-abstractive.html
2. Improved Paraphrase dataset and module, https://malaya.readthedocs.io/en/latest/load-abstractive.html
3. Improved Knowledge Graph triplet dataset and module, https://malaya.readthedocs.io/en/latest/load-knowledge-graph-triplet.html
4. added T5 for Knowledge Graph triplet, now able to predict for news, https://malaya.readthedocs.io/en/latest/load-knowledge-graph-triplet.html
5. Remove all t2t models, and replaced with T5-BASE, T5-SMALL and T5-TINY.
6. Now able to parallelise T5 models while maintaining model definition from Tensorflow-Mesh.