Gravity from Entropy theory offers a new resolution to one of physics' oldest paradoxes: how the universe builds galaxies and ...

A new study by Queen Mary University of London mathematician Professor Ginestra Bianconi proposes a new perspective on one of ...

PLACE 20 coins, heads up, on a tray and film it as you give it a shake. Then play the film backwards. From a jumbled mess, the coins all jump and come to rest with the same side up – an unreal, ...

A new theoretical study offers a possible explanation for how the Universe can grow more complex without violating the second ...

When two black holes orbit each other, they eventually spiral inward and collide in one of the most violent phenomena in the ...

Exactly 200 years ago, a French engineer introduced an idea that would quantify the universe’s inexorable slide into decay. But entropy, as it’s currently understood, is less a fact about the world ...

基于可验证奖励的强化学习（Reinforcement Learning with Verifiable Rewards，RLVR）正在成为大模型后训练的关键技术。数学题能判对错，代码能跑测试，可验证奖励让大模型可以通过强化学习持续提升推理能力。不过，在 RLVR 训练过程中，一个值得关注的问题是：模型并不总是越训练越会探索。随着策略熵（policy entropy）快速下降，模型输出可能逐渐同质化 ...