Model request: RecurrentGemma

[redbrain]

Model description

RecurrentGemma is a family of open language models built on Griffin, a novel recurrent architecture developed at Google. Both pre-trained and instruction-tuned versions are available in English.

Like Gemma, RecurrentGemma models are well-suited for a variety of text generation tasks, including question answering, summarization, and reasoning. Because of its novel architecture, RecurrentGemma requires less memory than Gemma and achieves faster inference when generating long sequences.

RecurrentGemma is already in transformers; it just lacks the features that TGI's officially supported models have (ie sharding).

Open source status

• The model implementation is available
• The model weights are available

Provide useful links for the implementation

https://ai.google.dev/gemma/docs/recurrentgemma/model_card
https://huggingface.co/google/recurrentgemma-2b-it
https://www.kaggle.com/models/google/recurrentgemma
https://github.com/huggingface/transformers/tree/main/src/transformers/models/recurrent_gemma
https://github.com/google/gemma.cpp