Together AI is a cloud platform for building and running generative AI models. The platform offers the fastest cloud for generating AI, built on leading research innovations from its team behind breakthrough models, datasets, and optimizations. Together's research includes Cocktail SGD which addresses key hindrances to training generative AI models in distributed environments by reducing network overhead by up to 117x. FlashAttention-2 accelerates training fine-tuning of LLMS by up to 9x and achieves an unprecedented 72% model flops utilization for training on Nvidia A100s. Together has partnered with Hazy Research to release Monarch Mixer - an alternative more efficient model architecture that enables scaling more efficiently while still matching quality.

Actionable recommendations:
1) Use Together Inference: Start using their easy-to-use API immediately benefitting from the fastest backend optimizations anywhere.
2) Fine-tune open-source models with your data using Together Fine-Tuning: Control hyperparameters like learning rate, batch size & epochs.
3) Get your own private GPU cluster with optimized distributed high-performance computing over fast infiniband networks via "Together GPU Clusters."
4) Build custom state-of-the-art Generative-AI Models From Scratch With Ease Via "Together Custom Models."

‍