GPT-2 for Tamil
A GPT-2 language model pretrained from scratch for Tamil, built during the Flax/JAX community week organised by HuggingFace.
Abstract
This repository was created as part of the Flax/JAX community week by HuggingFace. The aim of this project is to pretrain a language model using GPT-2 specifically for the Tamil language, using a causal language modeling (CLM) objective. The pretrained model is published on the HuggingFace Hub, and the accompanying space demonstrates it alongside other Tamil language models.
Link to model: abinayam/gpt-2-tamil
Link to demo: Tamil Language Demos