Evie Hu

on the atlas — 37

• Turns Out, You Can’t Just Bomb a Datacenter - by Sophie Kim2 savers
• Is there a Half-Life for the Success Rates of AI Agents? — Toby Ord1 savers
• Learning_curve1 savers
• [1709.06560] Deep Reinforcement Learning that Matters1 savers
• [2506.17209] Fine-Tuning Lowers Safety and Disrupts Evaluation Consistency1 savers
• They're Made Out of Weights1 savers
• Unfalsifiable stories of peace | Mechanize War, Inc.1 savers
• Over- and under-diagnosis of sexism • Otherwise1 savers
• Where is AI in GDP statistics? | PIIE1 savers
• List_of_constructed_languages1 savers
• Paper Trails1 savers
• A Mad Day’s Work: From Grothendieck to Connes and Kontsevich the Evolution of Concepts of Space and Symmetry1 savers
• Exploit Evals \ red.anthropic.com1 savers
• everblue notes1 savers
• ‘I didn't like certain aspects’: Trump postpones AI executive order - POLITICO1 savers
• Economy of Dreams: Hope in Global Capitalism and Its Critiques - Miyazaki - 2006 - Cultural Anthropology1 savers
• [2605.13829] Negation Neglect: When models fail to learn negations in training1 savers
• Curius / Onboarding2507 savers
• The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.34 savers
• Half-assing it with everything you've got30 savers
• The Persona Selection Model: Why AI Assistants might Behave like Humans28 savers
• The Future Worth Building Is Human - Thinking Machines Lab19 savers
• Verbalizable Representations Form a Global Workspace in Language Models17 savers
• The 37 Implementation Details of Proximal Policy Optimization · The ICLR Blog Track9 savers
• MATS 9 Retrospective & Advice — LessWrong8 savers
• My picture of the present in AI — LessWrong8 savers
• Off Target | CNAS7 savers
• Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations6 savers
• Built to benefit everyone: our plan | OpenAI6 savers
• Map of my Internet6 savers
• The Stanford Freshmen Who Think They Rule the World - The Atlantic6 savers
• Mechanize Inc.4 savers
• Plans A, B, C, and D for misalignment risk4 savers
• Should We Train Against (CoT) Monitors? — LessWrong3 savers
• [2502.17424] Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs3 savers
• Self-Serving utilitarian arguments2 savers
• They're Made out of Meat | East of the Web2 savers