Get the latest tech news
Implementation of Google's Griffin Architecture – RNN LLM
Open weights language model from Google DeepMind, based on Griffin. - google-deepmind/recurrentgemma
RecurrentGemma is a family of open-weights Language Models by Google DeepMind, based on the novel Griffin architecture. To run these notebooks you will need to have a Kaggle account and first read and accept the Gemma license terms and conditions from the RecurrentGemma page. The code has been optimized for running on TPU using the Flax implementation, which contains a low level Pallas kernel to perform the linear scan in the recurrent layers.
Or read this on Hacker News
