Brief

Several AI infrastructure companies have achieved significant funding milestones, with Turbopuffer reaching $100M ARR and profitability, Exa securing $250M in a Series C round valuing it at $2.2B, and Modal raising $355M at a $4.7B valuation. The AI News digest also highlighted advancements in model research, including RAEv2 for unified vision understanding and generation, NVIDIA's Gated DeltaNet-2 for improved language modeling, and a study questioning the necessity of subword tokenization. Additionally, discussions touched upon mechanistic interpretability and the potential for AI to drive breakthroughs in mathematics research, though with some skepticism regarding specific claims. AI

Researchers have introduced a new parameter-free method called "aligned training" to enhance the quality and stability of sparse autoencoders (SAEs), a technique used for interpreting deep neural networks. This method addresses issues like unused features and instability without requiring additional data or complex training procedures. Separately, a new approach called RAEv2 has been developed to improve Representation Autoencoders (RAEs), which are used in conjunction with pre-trained vision encoders. RAEv2 simplifies design choices and achieves state-of-the-art results in image generation tasks with significantly faster convergence. AI