Aya Expanse

The next generation of our open source large language model. This release includes model weights and starting code for pretrained and fine-tuned Llama language models — ranging from 7B to 70B parameters. Llama 2 pretrained models are trained on 2 trillion tokens, and have double the context length than Llama 1. Its fine-tuned models have been trained on over 1 million human annotations. Llama 2 outperforms other open source language models on many external benchmarks, including reasoning, coding, proficiency, and knowledge tests. Llama 2 was pretrained on publicly available online data sources. The fine-tuned model, Llama-2-chat, leverages publicly available instruction datasets and over 1 million human annotations. We have a broad range of supporters around the world who believe in our open approach to today’s AI — companies that have given early feedback and are excited to build with Llama 2.

NVIDIA Parakeet-RNNT-1.1B is a multilingual automatic speech recognition model built for quality transcription across voice applications. With 1.1 billion parameters and training on more than 90,000 hours of speech, it supports 25 languages and regional variants, including English, Spanish, French, German, Italian, Arabic, Japanese, Korean, Portuguese, Russian, Hindi, Dutch, Danish, Norwegian, Czech, Polish, Swedish, Thai, Turkish, and Hebrew. The model automatically detects the spoken language and uses a universal tokenizer created by training language-specific tokenizers and merging them into a shared vocabulary, enabling efficient cross-lingual learning and deployment. Parakeet-RNNT produces case-sensitive transcripts with upper and lowercase text, punctuation, spaces, and apostrophes, making the output suitable for production voice applications and downstream language understanding.