Now here's a cool use for AI!
"DeepMind’s GraphCast and WindBorne’s WeatherMesh can both run on consumer-grade computers, and can complete in a matter of seconds work that would usually require a supercomputer under traditional models."
https://www.semafor.com/article/02/14/2024/windborne-takes-the-ai-weather-prediction-crown
Learned classifiers should often possess certain invariance properties meant to encourage fairness, robustness, or out-of-distribution generalization. However, multiple recent works empirically demonstrate that common invariance-inducing regularizers are ineffective in the over-parameterized regime, in which classifiers perfectly fit (i.e. interpolate) the training data. This suggests that the phenomenon of "benign overfitting", in which models generalize well despite interpolating, might not favorably extend to settings in which robustness or fairness are desirable. In this work we provide a theoretical justification for these observations. We prove that -- even in the simplest of settings -- any interpolating learning rule (with arbitrarily small margin) will not satisfy these invariance properties. We then propose and analyze an algorithm that -- in the same setting -- successfully learns a non-interpolating classifier that is provably invariant. We validate our theoretical observations on simulated data and the Waterbirds dataset.
Fascinating #DeepLearninig paper about how #LargeLanguageModel exhibit new (emergent) capabilities when they scale up in size. It seems that as the models get bigger they start to be able to tackle new problems that they weren't explicitly trained on (e.g. arithmetic).
Scaling up language models has been shown to predictably improve performance and sample efficiency on a wide range of downstream tasks. This paper instead discusses an unpredictable phenomenon that we refer to as emergent abilities of large language models. We consider an ability to be emergent if it is not present in smaller models but is present in larger models. Thus, emergent abilities cannot be predicted simply by extrapolating the performance of smaller models. The existence of such emergence implies that additional scaling could further expand the range of capabilities of language models.
I'm so fascinated by the outcome of deep learning hyper networks, even a small 5k train in 256x256 gave me that result.
Tried to create some #cyberpunk inspired magazine covers with #DeepLearninig a bit of tweaking gave good surprisingly good results.
More at this insta account (will try to setup a better alternative next days) https://www.instagram.com/p/CkWshquoZdQ/?igshid=YmMyMTA2M2Y=
Andrew
Dennis Alexis Valin Dittrich
Sergrass 🇺🇦
Martin Schrimpf
Paul M. Heider
The Data Therapist
.:M:G:.