[{"@context":"https:\/\/schema.org\/","@type":"BlogPosting","@id":"https:\/\/blog.terabox.com\/insights\/grokking-ia-interpretacion-mecanicista#BlogPosting","mainEntityOfPage":"https:\/\/blog.terabox.com\/insights\/grokking-ia-interpretacion-mecanicista","headline":"Grokking: \u00bfPor qu\u00e9 la IA aprende de repente a generalizar?","name":"Grokking: \u00bfPor qu\u00e9 la IA aprende de repente a generalizar?","description":"\ud83d\udcfa V\u00eddeo de estudio recomendado hoy: https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4 \u00bfPor qu\u00e9 la IA “entiende” de repente? Descifrando el misterio del GrokkingEl misterio del “clic” algor\u00edtmicoLa geometr\u00eda oculta: Rotaciones y c\u00edrculosMedidas de progreso: Visualizando lo invisibleConclusiones clavePreguntas y Respuestas \u00bfPor qu\u00e9 la IA... ","datePublished":"2026-08-18","dateModified":"2026-08-18","author":{"@type":"Person","@id":"https:\/\/blog.terabox.com\/author\/flextech-admin\/#Person","name":"flextech-admin","url":"https:\/\/blog.terabox.com\/author\/flextech-admin\/","image":{"@type":"ImageObject","@id":"https:\/\/secure.gravatar.com\/avatar\/ad516503a11cd5ca435acc9bb6523536?s=150&d=mm&r=gforcedefault=1","url":"https:\/\/secure.gravatar.com\/avatar\/ad516503a11cd5ca435acc9bb6523536?s=150&d=mm&r=gforcedefault=1","height":96,"width":96}},"publisher":{"@type":"Organization","name":"terabox","logo":{"@type":"ImageObject","@id":"http:\/\/blog.terabox.com\/wp-content\/uploads\/2021\/11\/logo\u4ea7\u54c1\u540d-\u7ad6\u7248.png","url":"http:\/\/blog.terabox.com\/wp-content\/uploads\/2021\/11\/logo\u4ea7\u54c1\u540d-\u7ad6\u7248.png","width":900,"height":900}},"image":{"@type":"ImageObject","@id":"https:\/\/img.youtube.com\/vi\/IHikLL8ULa4\/maxresdefault.jpg","url":"https:\/\/img.youtube.com\/vi\/IHikLL8ULa4\/maxresdefault.jpg","height":"","width":""},"url":"https:\/\/blog.terabox.com\/insights\/grokking-ia-interpretacion-mecanicista","video":{"@context":"http:\/\/schema.org\/","@type":"VideoObject","@id":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4#VideoObject","contentUrl":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4","name":"A Walkthrough of Progress Measures for Grokking via Mechanistic Interpretability: What? (Part 1\/3)","description":"Part 1 of a walkthrough of our paper, Progress Measures for Grokking via Mechanistic Interpretability. I'm joined by my co-author Lawrence Chan. In this part, we give an overview of the paper and discuss the key takeaways\n\nPart 2: https:\/\/youtu.be\/C6Lil5CiuMI\nPart 3: https:\/\/youtu.be\/-zRx0vMpLl4\n\nIf you want to learn more about mechanistic interpretability, check out https:\/\/neelnanda.io\/getting-started\n\nOur paper: https:\/\/arxiv.org\/abs\/2301.05217\nOriginal grokking paper: https:\/\/arxiv.org\/abs\/2201.02177\nAdamW: https:\/\/pytorch.org\/docs\/stable\/generated\/torch.optim.AdamW.html\nWalkthrough of toy models of superposition: https:\/\/www.youtube.com\/watch?v=R3nbXgMnVqQ\nDanny Hernandez paper on scaling laws for repeated data: https:\/\/arxiv.org\/abs\/2205.10487\nJermyn & Schlegeris on S-Shaped Curves: https:\/\/www.alignmentforum.org\/posts\/RKDQCB6smLWgs2Mhr\/multi-component-learning-and-s-curves\nUnifying Grokking and Double Descent: https:\/\/arxiv.org\/abs\/2303.06173\nOmnigrok: https:\/\/arxiv.org\/abs\/2210.01117\n\n0:00 - Intro\n0:50 - What is grokking? \n9:53 - Mechanistic interpretability\n11:47 - Paper overview, modular addition algorithm\n15:08 - Progress measures\n21:41 - why this work is bullshit\n29:30 - Predicting when it will grok?\n33:45 - Why does grokking happen?\n40:27 - Lottery ticket hypothesis\n42:43 - Conclusion","thumbnailUrl":["https:\/\/i.ytimg.com\/vi\/IHikLL8ULa4\/default.jpg","https:\/\/i.ytimg.com\/vi\/IHikLL8ULa4\/mqdefault.jpg","https:\/\/i.ytimg.com\/vi\/IHikLL8ULa4\/hqdefault.jpg","https:\/\/i.ytimg.com\/vi\/IHikLL8ULa4\/sddefault.jpg","https:\/\/i.ytimg.com\/vi\/IHikLL8ULa4\/maxresdefault.jpg"],"uploadDate":"2023-04-10T14:36:00+00:00","duration":"PT43M10S","embedUrl":"https:\/\/www.youtube.com\/embed\/IHikLL8ULa4","publisher":{"@type":"Organization","@id":"https:\/\/www.youtube.com\/channel\/UCBMJ0D-omcRay8dh4QT0doQ#Organization","url":"https:\/\/www.youtube.com\/channel\/UCBMJ0D-omcRay8dh4QT0doQ","name":"Neel Nanda","description":"","logo":{"url":"https:\/\/yt3.ggpht.com\/Syi-a2wpm7bUZMpIqufmxgoAqhvVU_BD5OQg3dxsuU1Fu8suHtSrKunj3Po48uH6L03Z0MmeylE=s800-c-k-c0x00ffffff-no-rj","width":800,"height":800,"@type":"ImageObject","@id":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4#VideoObject_publisher_logo_ImageObject"}},"potentialAction":{"@type":"SeekToAction","@id":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4#VideoObject_potentialAction","target":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4&t={seek_to_second_number}","startOffset-input":"required name=seek_to_second_number"},"interactionStatistic":[[{"@type":"InteractionCounter","@id":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4#VideoObject_interactionStatistic_WatchAction","interactionType":{"@type":"WatchAction"},"userInteractionCount":7912}],{"@type":"InteractionCounter","@id":"https:\/\/www.youtube.com\/watch?v=IHikLL8ULa4#VideoObject_interactionStatistic_LikeAction","interactionType":{"@type":"LikeAction"},"userInteractionCount":214}]},"about":["\u300eSpanish\u300f","Insights"],"wordCount":1730},{"@context":"https:\/\/schema.org\/","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Insights","item":"https:\/\/blog.terabox.com\/insights\/#breadcrumbitem"},{"@type":"ListItem","position":2,"name":"Grokking: \u00bfPor qu\u00e9 la IA aprende de repente a generalizar?","item":"https:\/\/blog.terabox.com\/insights\/grokking-ia-interpretacion-mecanicista#breadcrumbitem"}]}]