Sarvam 30B and Sarvam 105B are open-source reasoning models trained from scratch on large-scale, high-quality datasets. The training was conducted in India under the IndiaAI mission, optimizing various aspects including tokenization, model architecture, and execution kernels.

Sam Altman emphasizes that training an AI model requires significant energy, comparable to the 20 years and nutrition needed for human intelligence development. Demis Hassabis suggests testing AI by training it with a knowledge cutoff of 1911 to evaluate its ability to derive concepts like general relativity independently.

Reinforcement learning from human feedback (RLHF) is a key technique for deploying advanced machine learning systems. A new book provides an introduction to the core methods of RLHF for readers with a quantitative background.