Google unveils open supply initiatives for generative AI

Google at Google Cloud Subsequent 24 unveiled three open supply initiatives for constructing and working generative AI fashions. The corporate additionally launched new giant language fashions to its MaxText venture of JAX-built LLMs.

The brand new LLM fashions in MaxText embody Gemma, GPT-3, Llama 2, and Mistral, that are supported throughout each Google Cloud TPUs and Nvidia GPUs, the corporate mentioned. 

The newly unveiled open supply initiatives are MaxDiffusion, JetStream, and Optimum-TPU.

MaxDiffusion is a set of high-performance and scalable reference implementations for diffusion fashions comparable to Secure Diffusion. Just like the MaxText fashions, the MaxDiffusion fashions are constructed on JAX, which is a framework for high-performance numerical computing and large-scale machine studying. 

JAX in flip is built-in with the OpenXLA compiler, which optimizes numerical capabilities and delivers wonderful efficiency at scale, permitting mannequin builders to concentrate on the maths and let the software program drive the best implementation. 

“We’ve closely optimized JAX and OpenXLA efficiency on Cloud TPU and partnered intently with Nvidia to optimize OpenXLA efficiency on giant Cloud GPU clusters,” Google mentioned.

The corporate additionally launched Jetstream, which is an open supply optimized LLM inference engine supporting XLA compilers. 

“As prospects convey their AI workloads to manufacturing, there’s an rising demand for a cost-efficient inference stack that delivers excessive efficiency. JetStream helps with this want and presents help for fashions educated with each JAX and PyTorch/XLA, and contains optimizations for widespread open fashions comparable to Llama 2 and Gemma,” Mark Lohmeyer, normal supervisor of compute and ML infrastructure at Google Cloud, mentioned.

Lastly, Google’s open supply bulletins included the launch of Optimum-TPU for PyTorch customers within the Hugging Face neighborhood. Optimum-TPU brings Google Cloud TPU efficiency optimizations for each coaching and inference. It helps the Gemma 2b mannequin now and Llama and Mistral quickly, Google mentioned.