Research

Superposition, memorization, and double descent

Anthropic

Toy models show networks store data points in superposition when data is scarce, switching to generalizable features as data grows.

Visit Site

Research Anthropic

ResearchConstitutional Classifiers: Defending against universal jailbreaksAnthropic ResearchAuditing language models for hidden objectivesAnthropic ResearchEnabling independent research on how people use ClaudeAnthropic ResearchProject Swap: What happens when agents trade for us?Anthropic ResearchForecasting rare language model behaviorsAnthropic ResearchDisempowerment patterns in real-world AI usageAnthropic BlogsRocketCDN Partners with Bunny.net to Power 12,000+ WebsitesBunny BlogsThe Double-Edged Sustainability Sword Of AI In Web DesignSmashingmagazine BlogsAI adoption starts with truthReplit Blogs30 to 70 PRs a Day: How We Managed to Not Wreck Our SystemsHoneycomb BlogsCTO Chat: Overcoming the Developer Experience Gap (feat. RedMonk & Flow.io)Docker BlogsComparing Native Postgres, ElasticSearch, and pg_search for Full-Text Search - NeonNeon BlogsGroq LPU Tops Latency & Throughput in BenchmarkGroq ResearchDistributed representations: Composition and superpositionAnthropic ResearchToy models of superpositionAnthropic Products & ServicesType-C to Type-C PD Fast Charging Cabledfrobot.com Products & ServicesSipeed Maix Go Development Kit for RISC-V AIdfrobot.com Products & ServicesExtruder 50mm Fandfrobot.com ShoppingFrameworkframe.work Products & Services6xAA Battery Holder Double Layer with Leadsdfrobot.com