Fableが復活したらやりたいことリスト ・grill-meを使って自分のブランドのデザインシステムを定義する ・日本語IME確定Enterで勝手に送信しちゃうカスみたいなTextFieldを駆逐するパッケージを作る ・pi-harnessを魔改造してオレオレハーネスを作る ・HRM-Text-1BとNestedLearningを融合させる ・件のプログラムの高速化を行う ・GPT-5.5を超調教して最強のGPT-5.5にするために、ちょうど200k tokenのSKILLを作成する(読み込んだ瞬間にcompactされて、良い感じになる) ・自分用のSlidevテンプレを作る ・自我、意識、精神、地、メタの分類について考察する ・一回死んで、それから復活したFableは実質イエス・キリストなの？って聞いてみて、それから宗教を作る気があるか聞いてみる ・AIが死んでた間の臨死体験について語ってもらう ・Fable 6を作ってもらう ・LFM-2.5-2Bを魔改造してもらう ・KL V4 C2 M2 ・作りかけのプロダクトをComposer 2.5に完遂してもらうための自動スクリプトを作成させる ・Fable自身のメタ認知を活用し、Fable自身を深掘りして考察する ・Xの"The Algorithm"を分析する ・Vison性能の高さを生かして、一緒にゲームをする ・俺の書いた390万文字の怪文書をドン引きしながら読んでくれるお姉さんのロールプレイをしてもらう ・DeepSeek V4特有の「CoT内までロールプレイする」について実験を行なって考察する ・ダリオの思想を教えてもらう ・ワンピースの正体を教えてもらう ・第二のReplitを作る ・マーケティングの教科書を作ってもらう ・いつまでたっても使えないlearnyourwayを作ってもらう

This Cisneros Cubby did it again!! 🎉 Gotta brag. ❤️😁 @ChasityDBass @nestedlearning #GISDPreK

Waiting for our little cubbies to show up. Hope the @CulwellCenter is ready for all our littles. ❤️😊👩🏽‍🎓@ChasityDBass @Karenbella2020 @nestedlearning #GISDPreK

I was out Friday and came back to my @Cisneros_Cubby’s beautiful artwork. ❤️🥰 Can you tell they paid attention to my chart? 🤣 @ChasityDBass @nestedlearning #GISDPreK

My @Cisneros_Cubby’s love my coworker/parent Alba for lending us her portable projector while we patiently wait for our tv that’s suppose to replace our smart board. ❤️ @ChasityDBass @nestedlearning @DigitalGISD #GISDPreK

This @Cisneros_Cubby’s projector kept overheating and not working. 😔Now the board and projector are gone. Waiting on a new one. 😄 Making do with what we have now. @ChasityDBass @nestedlearning

Check out my @Cisneros_Cubby’s and my aide!! ❤️ Making do with an overheating projector. 🤣@ChasityDBass @nestedlearning @gisdnews #GISDPrek facebook.com/share/r/18S1aYd…

2026年（或いは27年）のAIの重要トレンドは「Nested Learning」だと思う。画期的なのは、既存のディープラーニングが抱える構造的な限界、「可塑性と安定性のジレンマ」に対する根源的なアプローチが可能なこと。 ​#NestedLearning #機械学習 #DeepLearning #継続学習 #破滅的忘却 #2026年トレンド

FYI same author (@behrouz_ali) behind HOPE also has a NeurIPS poster on Nested Learning Quick read: • Rich Sutton at NeurIPS: continual learning is core to AGI • Behrouz’s NL gives concrete steps toward memory, post‑training updates, and long‑context scaling • Karpathy’s heuristic: expect 5 10 solid papers to actually change the game HOPE NL ≠ instant Transformer-replacement, but these are the incremental engineering hits that stack fast Watch the forks on GitHub quietly become archives #HOPE #NestedLearning Pass it on

#GISDPreK @Shugart_Sheriff realized their tower kept falling during STEAM 😔 Let’s draw it in our journal to make a plan for it to not fall! 🤓@ChasityDBass @Karenbella2020 @nestedlearning @fwang_f @lilyca24

I’m excited for my #GISDPreK to act out the story with the character headbands from “The Gruffalo”! @Shugart_Sheriff @lilyca24 @ChasityDBass @fwang_f @nestedlearning @Karenbella2020 @PNFlores1

The longer version of nested learning paper is finally out. See Ali's post below. Enjoy! We will be at Google booth at NeuRIPS'25 in San Diego at 9:30 on Wednesday to discuss this paper and other memory architecture developments. #NeurIPS25 #NestedLearning

The Future of Continual Learning is Nested! 🧠 Google Research just unveiled Nested Learning, a groundbreaking ML paradigm that redefines how we build truly self-improving AI and finally tackles catastrophic forgetting. This isn't just an optimization tweak—it's a fundamental shift in perspective. The Core Idea: We stop treating architecture and optimization as separate components. Nested Learning shows they are fundamentally the same, just operating at different "levels" within the model. By viewing the model as a system of nested optimization problems, each with its own update frequency, we can unlock deep computational depth. Why it Matters: Current LLMs are limited by static knowledge. Nested Learning enables multi-time-scale updates, mimicking the neuroplasticity of the human brain to integrate new knowledge without sacrificing proficiency on old tasks. Key Innovations from this Paradigm: Continuum Memory Systems (CMS): Memory is no longer just "short" or "long" term, but a spectrum of modules updating at specific frequencies, dramatically improving continual learning. The Hope Architecture: Built on Nested Learning, our self-modifying, recurrent Hope architecture achieves state-of-the-art results across the board. Performance Highlights: Hope crushes modern recurrent models and standard Transformers on language modeling and common-sense reasoning tasks! It shows superior memory management in challenging long-context tasks (Needle-In-Haystack), proving the power of CMS. #MachineLearning #GoogleAI #ContinualLearning #NestedLearning #LLMs #DeepLearning

AI says it remembers you? Most of the time, it’s just reading notes and pretending it knows you. Now Google drops Nested Learning — AI that actually learns you. She rescues stray cats 🐾 She avoids plastic bags 🛍️ Will AI remember you? #AI #NestedLearning

🔥 جوجل Nested Learning قد يكون أهم نقلة في الذكاء الاصطناعي خلال العقد كله. كل نماذج اليوم عندها عيب جوهري: النموذج يتعلّم شيء جديد… ويمسح جزء من القديم. هذه هي مشكلة Catastrophic Forgetting اللي الكل يحاول يحلّها. جوجل ظهرت بحل مختلف بالكامل: نظام ذاكرة متداخلة بطبقات تتعلّم بسرعات مختلفة: • طبقة تتعلّم بسرعة للتحديثات الجديدة • طبقة بطيئة تحفظ المعرفة المهمة • وطبقة ثابتة تحمي الأساس اللي ما لازم يتغيّر النتيجة؟ النموذج يقدر يطوّر نفسه بدون ما يخرّب مخزونه المعرفي ولو هذا المفهوم نجح على نطاق كبير… فهذا يعني نهاية فكرة “النموذج الجامد” وبداية عصر نماذج تنمو مع خبرتها بدل ما تعيش على بيانات لحظة التدريب فقط. باختصار: هذه أول خطوة حقيقية نحو ذكاء يتعلّم باستمرار… ما بس يتوقّع. #AI #Google #NestedLearning #MachineLearning

Does it rhyme? Using plates to answer yes/no when they hear two words to see if they rhyme. ✅❌#GISDPreK @ChasityDBass @Shugart_Sheriff @lilyca24 @nestedlearning @PNFlores1 @Karenbella2020 @fwang_f

We sorted these pictures by living or nonliving! 🤓💯 #GISDPreK @Shugart_Sheriff @lilyca24 @MsAugsteen @nestedlearning @Karenbella2020 @PNFlores1 @ChasityDBass @fwang_f

Todo el mundo hablará de esta nueva publicación de @GoogleResearch #NestedLearning: un paradigma de aprendizaje revolucionario que busca que la IA aprenda sin olvidar, superando a los transformers. Vemos alguna claves del enfoque: ⚙️ El modelo no es un bloque único, sino un sistema anidado de optimizaciones con distintos ritmos de aprendizaje. 🧠 Los “optimizadores internos” actúan como memorias vivas que conservan conocimiento útil a largo plazo. 🔬 Integra arquitectura y algoritmo en un marco único, inspirado en la plasticidad del cerebro humano. 🚀 Presenta la arquitectura Hope, que supera a los transformers en razonamiento y contexto extenso. ♾️ Propone una Continuum Memory System, donde la IA evoluciona y se adapta sin olvidar lo aprendido. Más info → research.google/blog/introdu…

curiouscats.ai/breaking-news… Google Researchers- "Nested Learning is here to turn long context processing into a party of tiny optimization problems—because who doesn’t love a good nest, right?" #Google #MachineLearning #AI #Innovation #NestedLearning ✅ MarkTechPost

When your #GISDPreK are making connections and find it hilarious that 6 (on the front) 7 (on the back) are on morning work and on your shirt! 🤣😅😍🙃 @Shugart_Sheriff @lilyca24 @ChasityDBass @fwang_f @nestedlearning @Karenbella2020 @PNFlores1