First Modern Backpropagation ​

In his 1976 paper “Taylor Expansion of the Accumulated Rounding Error,” S. Linnainmaa presented a novel approach for addressing rounding errors in numerical calculations through the use of Taylor series expansion. By systematically expanding the rounding...

Gemini

The Gemini paper by Google DeepMind presents advanced multimodal AI models capable of integrating and processing text, images, audio, and video within a single framework. The initial model, Gemini 1.0, is offered in three sizes—Ultra, Pro, and Nano—each designed for...

Human or Not

An online game inspired by the Turing test, that measures the capability of AI chatbots to mimic humans in dialog, and of humans to tell bots from other humans. Over the course of a month, the game was played by over 1.5 million users who engaged in anonymous...

Llama

The LLaMA (Large Language Model Meta AI) paper describes a family of large language models developed by Meta AI to provide open and efficient foundation models for various natural language processing tasks. The LLaMA models use techniques like RMSNorm, SwiGLU...