Together AI Core Team Bridge the Gap Between AI and Hardware

Vad har hänt

Together AI has spotlighted its internal core team, whose primary task is to develop high-performance and optimised software components known as kernels. These kernels are crucial for maximising efficiency when running AI models on graphics processing units (GPUs). The team is behind the successful FlashAttention architecture and the ThunderKittens optimisation library.

Varför det spelar roll

The development of optimised kernels is critical to meeting the increasing demands for computational power in the AI sector. By bridging the gap between software and hardware, Together AI can offer faster and more cost-effective solutions for AI development. This accelerates the pace of innovation and makes advanced AI more accessible.

Vem påverkas

Primarily AI developers and companies using or building large-scale AI models are affected. Cloud providers and hardware manufacturers such as NVIDIA are also impacted, as optimised kernels push the limits of hardware and drive development forward. Indirectly, end-users of AI services benefit from faster and more responsive applications.

EU-status

Ej relevant för EU-status.

Mer att veta

Work on kernel optimisation requires in-depth knowledge of both algorithms and low-level programming for specific hardware, making these teams a vital strategic asset for AI companies.