# Large-scale Corpus Training

A large Japanese RoBERTa model pretrained on Japanese Wikipedia and the Japanese portion of CC-100, suitable for Japanese natural language processing tasks.

This is a neural machine translation model for translating from Catalan, Occitan, and Spanish to English, part of the OPUS-MT project.

This is a neural machine translation model for translating from English to Arabic, part of the OPUS-MT project, supporting multi-target language translation.

An Icelandic masked language model based on the RoBERTa-base architecture, fine-tuned from xlm-roberta-base

Icelandic masked language model trained on RoBERTa-base architecture using the fairseq framework

Portuguese BERT model fine-tuned for MLM (Masked Language Modeling) on 500,000 instances from the Brazilian Federal Official Gazette, based on the Bertimbau-Base model

GerPT2 is the large-scale version of the German GPT2, trained on the CC-100 corpus and German Wikipedia, excelling in German text generation tasks.

QARiB is a BERT model based on Arabic and its dialects, trained on approximately 420 million tweets and 180 million text sentences, supporting various Arabic NLP tasks.

QARiB is an Arabic and dialect BERT model trained on approximately 420 million tweets and 180 million text sentences, supporting various downstream NLP tasks.

T5 (Text-to-Text Transfer Transformer) base model pretrained on Indonesian mC4 dataset, requires fine-tuning for use

Russian conversational model trained on OpenSubtitles, Dirty, Pikabu, and Taiga corpus social media sections

RoBERTa model trained on a 43GB dataset of Croatian and Serbian languages, supporting masked language modeling tasks.

Est-RoBERTa is a monolingual Estonian BERT-like model based on the RoBERTa architecture, trained on 2.51 billion Estonian vocabulary tokens.