#foundationmodel — Public Fediverse posts

RE: https://mastodon.social/@thetransmitter/116476879384094089

Tl;dr: transformers didn’t learn the underlying principle.

“Keyon Vafa and his colleagues simulated tens of millions of planetary orbits from Newton’s laws and trained a transformer on the resulting sequences. The model predicted future positions with very high accuracy. But when they fine-tuned it to infer the underlying gravitational force vectors, the model produced nonsense. The implied laws of gravitation were different depending on which subset of data the researchers examined. The transformer had assembled a patchwork of heuristics, accurate for every solar system in the training set, but it hadn’t discovered the universal gravitational principle. Without that principle, the model could predict the movement of points of light in the sky but could never send a rocket to the moon.”

Also link to the preprint: https://arxiv.org/abs/2507.06952

#neuroscience #prediction #FoundationModel

RE: https://mastodon.social/@thetransmitter/116476879384094089

Tl;dr: transformers didn’t learn the underlying principle.

“Keyon Vafa and his colleagues simulated tens of millions of planetary orbits from Newton’s laws and trained a transformer on the resulting sequences. The model predicted future positions with very high accuracy. But when they fine-tuned it to infer the underlying gravitational force vectors, the model produced nonsense. The implied laws of gravitation were different depending on which subset of data the researchers examined. The transformer had assembled a patchwork of heuristics, accurate for every solar system in the training set, but it hadn’t discovered the universal gravitational principle. Without that principle, the model could predict the movement of points of light in the sky but could never send a rocket to the moon.”

Also link to the preprint: https://arxiv.org/abs/2507.06952

#neuroscience #prediction #FoundationModel

RE: https://mastodon.social/@thetransmitter/116476879384094089

Tl;dr: transformers didn’t learn the underlying principle.

“Keyon Vafa and his colleagues simulated tens of millions of planetary orbits from Newton’s laws and trained a transformer on the resulting sequences. The model predicted future positions with very high accuracy. But when they fine-tuned it to infer the underlying gravitational force vectors, the model produced nonsense. The implied laws of gravitation were different depending on which subset of data the researchers examined. The transformer had assembled a patchwork of heuristics, accurate for every solar system in the training set, but it hadn’t discovered the universal gravitational principle. Without that principle, the model could predict the movement of points of light in the sky but could never send a rocket to the moon.”

Also link to the preprint: https://arxiv.org/abs/2507.06952

#neuroscience #prediction #FoundationModel

RE: https://mastodon.social/@thetransmitter/116476879384094089

Tl;dr: transformers didn’t learn the underlying principle.

“Keyon Vafa and his colleagues simulated tens of millions of planetary orbits from Newton’s laws and trained a transformer on the resulting sequences. The model predicted future positions with very high accuracy. But when they fine-tuned it to infer the underlying gravitational force vectors, the model produced nonsense. The implied laws of gravitation were different depending on which subset of data the researchers examined. The transformer had assembled a patchwork of heuristics, accurate for every solar system in the training set, but it hadn’t discovered the universal gravitational principle. Without that principle, the model could predict the movement of points of light in the sky but could never send a rocket to the moon.”

Also link to the preprint: https://arxiv.org/abs/2507.06952

#neuroscience #prediction #FoundationModel

New #FoundationModel reveals how #cells are organized in tissues. Developed by #HelmholtzMunich and #TUM.
https://nachrichten.idw-online.de/2025/10/30/new-foundation-model-reveals-how-cells-are-organized-in-tissues

Neues #FoundationModel zeigt, wie Zellen im Gewebe organisiert sind und miteinander interagieren. Entwickelt von #HelmholtzMunich und der #TUM.
https://nachrichten.idw-online.de/2025/10/30/neues-foundation-modell-zeigt-organisation-von-zellen-im-gewebe

Apple zeigt KI auf der WWDC - aber nur mit angezogener Handbremse

https://techupdate.io/apple/apple-zeigt-ki-auf-der-wwdc-aber-nur-mit-angezogener-handbremse/50416/

#siri #apple #ki #technews #ai #ios26 #macos26 #wwdc2025 #foundationmodel #appentwickler #gamecenter #macgaming

SK On plans to develop its own AI foundation model by 2028 to enhance competitiveness in battery manufacturing, integrating internal and external data across R&D, production, and strategic operations.
#YonhapInfomax #SKOn #ArtificialIntelligence #FoundationModel #BatteryManufacturing #RAndD #Economics #FinancialMarkets #Banking #Securities #Bonds #StockMarket
https://en.infomaxai.com/news/articleView.html?idxno=53030

Segment Anything Model Can Not Segment Anything - Assessing AI Foundation Model’s Generalizability In Permafrost Mapping
--
https://doi.org/10.3390/rs16050797 <-- shared paper
--
#GIS #spatial #mapping #remotesensing #foundationmodel #AI #artificialintelligence #zeroshot #segmentation #GeoAI #spatialanalysis #LargeLanguageModel #LLM #SAM #performance #metrics #permafrost #visionmodel #icewedge #Arctic #warming #climatechange #thawslumps #landform #terrainmapping #EuroCrops #agriculture

Just heard a podcast with the #ceo of #amazon #aws Matt Garman referring to "bedrock" models to what others refer to as a #foundationmodel . That's a big #middlefinger to #Stanford !

"justice is what the judge ate for breakfast" - #AI is what the #foundationmodel was trained by #ChatGPT #axelspringerverlag