# \#self-supervised
**URL:** https://www.artofsm.art/tag/self-supervised/313.md
[Latest](https://www.artofsm.art/latest.md) · [Categories](https://www.artofsm.art/categories.md) · [Tags](https://www.artofsm.art/tags.md)
---
## [A Bank Trained Its Own Foundation Model to Catch Fraud — Freddy Lecue, Wells Fargo](https://www.artofsm.art/t/a-bank-trained-its-own-foundation-model-to-catch-fraud-freddy-lecue-wells-fargo/26701)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\
**Replies:** 1\
**Last updated:** [23 October 2026 09:24 UTC](https://www.artofsm.art/t/a-bank-trained-its-own-foundation-model-to-catch-fraud-freddy-lecue-wells-fargo/26701 "2026-10-23T09:24:16Z")
Freddy Lecue from Wells Fargo discusses the creation of a custom billion-parameter foundation model built from scratch using transformer architecture to analyze vast amounts of structured financial transact…
---
## [Yann LeCun's Billion Dollar Bet](https://www.artofsm.art/t/yann-lecuns-billion-dollar-bet/18260)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\
**Replies:** 1\
**Last updated:** [2 May 2026 02:05 UTC](https://www.artofsm.art/t/yann-lecuns-billion-dollar-bet/18260 "2026-05-02T02:05:25Z")
Yann LeCun has invested a billion dollars in developing JEPA, an AI approach that focuses on learning internal representations through joint embedding predictive architectures to build world models capable …
---
## [AI is SO Smart, Why Are Its Internals 'Spaghetti'?](https://www.artofsm.art/t/ai-is-so-smart-why-are-its-internals-spaghetti/10597)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\
**Replies:** 1\
**Last updated:** [4 July 2025 21:08 UTC](https://www.artofsm.art/t/ai-is-so-smart-why-are-its-internals-spaghetti/10597 "2025-07-04T21:08:40Z")
The video argues that despite AI’s impressive outputs, its internal representations are chaotic and lack true understanding due to reliance on stochastic gradient descent, which produces fragmented “spaghet…
---
## [He did it, The AGI Architecture Finally Worked!](https://www.artofsm.art/t/he-did-it-the-agi-architecture-finally-worked/10018)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\
**Replies:** 2\
**Last updated:** [15 June 2025 15:27 UTC](https://www.artofsm.art/t/he-did-it-the-agi-architecture-finally-worked/10018 "2025-06-15T15:27:48Z")
The video highlights Yan Lan’s breakthrough AGI architecture using world models that enable AI to understand and predict abstract representations of reality, demonstrated by the VJ Jepa 2 model’s advanced v…