Just links

2022-05-27 23:46:35 Inverse Occam’s razor https://www.nature.com/articles/s41567-022-01575-2
https://arxiv.org/abs/2204.08284 Открыть/Комментировать

2022-05-27 21:34:13 Discovering Policies with DOMiNO: Diversity Optimization Maintaining Near Optimality https://arxiv.org/abs/2205.13521 Открыть/Комментировать

2022-05-26 23:05:47 I think this counterpoint still stands

On the Paradox of Learning to Reason from Data
https://twitter.com/HonghuaZhang2/status/1528963938825580544
https://arxiv.org/abs/2205.11502 Открыть/Комментировать

2022-05-26 23:05:16 Autoformalization with Large Language Models https://arxiv.org/abs/2205.12615 Открыть/Комментировать

2022-05-25 22:45:16 https://www.nasa.gov/press-release/nasa-selects-futuristic-space-technology-concepts-for-early-study Открыть/Комментировать

2022-05-25 16:28:23 https://twitter.com/preskill/status/1529120459056304128 Открыть/Комментировать

2022-05-25 09:45:24 First Contact: Unsupervised Human-Machine Co-Adaptation via Mutual Information Maximization https://sites.google.com/view/coadaptation
#rl #unsupervised Открыть/Комментировать

2022-05-25 07:51:21 Work on automated reasoning & formal maths from MetaAI:

"Our model is able to prove mathematical theorems in a fully automated way and significantly outperforms the [state of the art]..."

"The model learns from previous proof searches through online training, which allows it to generalize to mathematical domains far from the training distribution. In Metamath, our model proves more than 82% of theorems from a random split (compared to 56.5% for GPT-f)."

Paper: HyperTree Proof Search for Neural Theorem Proving https://arxiv.org/abs/2205.11491 Открыть/Комментировать

2022-05-24 15:53:46 Identifying the ν=5/2 topological order through charge transport measurements https://arxiv.org/abs/2106.07667
#physics #qhe Открыть/Комментировать

2022-05-24 11:55:39 The Brunerie number is -2 https://groups.google.com/g/homotopytypetheory/c/588SusPGMbo Открыть/Комментировать