RecurrentGemma: Moving Past Transformers for Efficient Open Language Models

arXiv:2404.07839v1 Announce Type: new
Abstract: We introduce RecurrentGemma, an open language model which uses Google's novel Griffin architecture. Griffin combines linear recurrences with local attention to achieve excellent performance on language. It has a fixed-sized state, which reduces memory use and enables efficient inference on long sequences. We provide a pre-trained model with 2B non-embedding parameters, and an instruction tuned variant. Both models achieve comparable performance to Gemma-2B despite being trained on fewer tokens.

abstract architecture arxiv attention cs.ai cs.cl cs.lg google griffin inference language language model language models linear local attention memory moving novel performance state transformers type