PulseAugur / Brief
EN
LIVE 08:33:29

Brief

last 24h
[1/1] 222 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. Materials AlphaFold is here! 40 industrial tasks achieve SOTA across the board, AI4S welcomes a major industry breakthrough

    A new materials science foundation model called MPA (Materials Property Axiom) has been developed by Deep Principle, utilizing a training methodology inspired by large language models. This approach, which includes a mid-training phase for developing "physical intuition" and a novel "Hybrid Readout" head, allows MPA to achieve state-of-the-art results on 40 real-world industrial material tasks. The model demonstrates superior performance, particularly in predicting properties of novel material structures, indicating a significant advancement in AI for materials science. AI

    IMPACT Establishes a new benchmark for AI in materials science, potentially accelerating discovery and development of new materials by improving predictive accuracy on real-world tasks.