mesolitica
/

nanot5-base-malaysian-cased

text2text-generation

text-generation-inference

Model card Files Files and versions

Pretrain BASE 512 masking context length T5 on Malaysian text

README at https://github.com/mesolitica/malaya/tree/5.1/pretrained-model/nanoT5

WandB, https://wandb.ai/huseinzol05/nanoT5-base

Downloads last month: 25

Safetensors

Model size

0.2B params

Tensor type

F32

·

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for mesolitica/nanot5-base-malaysian-cased

Finetunes

Collection including mesolitica/nanot5-base-malaysian-cased

Malaysian Seq2Seq

Trained on 17B tokens, 81GB of cleaned texts, able to understand standard Malay, local Malay, local Mandarin, Manglish, and local Tamil. • 8 items • Updated Jun 24, 2025