# Low-resource language adaptation

The Jais series are specialized bilingual large language models for Arabic processing, adapted for Arabic through pre-training based on the Llama-2 architecture

A human-aligned dialogue model trained on Russian and English, optimized using Direct Preference Optimization, supporting bilingual Russian-English conversations.

Llammas is an open-source conversational model adapted for Estonian based on the Llama-2-7B model, achieving cross-lingual knowledge transfer through two-stage training

A multilingual processing model supporting over 150 languages, suitable for various natural language processing tasks.

A general-purpose language model supporting over 200 languages, suitable for various natural language processing tasks.

A foundational speech recognition model pretrained on 10,000 hours of unlabeled data from the VoxPopuli corpus, supporting multilingual speech processing

The first monolingual Transformer model specifically designed for Spanish news abstract generation, enhancing abstractiveness through specialized pre-training, outperforming mainstream multilingual models.

This is a Transformer-based multilingual-to-English machine translation model supporting over 50 Indo-European languages.