Picks

The atlas

Every source, article and topic this publication shows, and how they connect.

a thing an ideaa ring for each source about it a source in the issue

#3-SATAcademic reward systemadversarial suffixAgent harnessAI adoptionAI-assisted mathematicsAI cybersecurityAI sovereigntyAMPS firewall paradoxApplied mathematicsArtificial general intelligenceArtificial intelligenceblack hole complementarityBlock-diagonal-plus-low-rank matrixClaude CodeCodexcoding agent setup with a coding agentconfidence buildup during loopsCordisData centerDeepSeekDeepSeek HarnessDeepSeek Sparse AttentionDeepSeek Super AgentDirectional ablationDoctoral educationdsh-superagentExponential Time Hypothesisfinite-time blowupFrontier AI modelFrontier Harness EvalGemma 3 270MGeometric complexity theoryGLM-4.7GLM-5GPT-6 Astragreedy decodingHawking radiation decodingHermite interpolationInference infrastructureknowledge distillationLow-rank matrixMATH-500Mathematical understandingMathematicsMathematics educationminimal human attention principleNavier–Stokes equationsNavier–Stokes existence and smoothnessOpenAIP versus NP problemParity functionPermanentPlugin architecturePolynomial approximation of Boolean functionsproactive coding agentProximal policy optimizationquantum error correctionQwen3.5-0.8Brefusal directionrefusal in large language modelsRepetitive reasoningrisk aversion from hardness of learningScientific authorshipself-similar collapsing vortex coreslimeSteven StrogatzSycophancytemperature (sampling)temporally correlated errorsweight orthogonalizationWeight-space programmingScott Aaronson: The Mathocalypse, Will Intelligence Matter Anymore?arxiv-0704.0229Black Holes as Mirrors: Quantum Information in Random Subsystemsarxiv-0709.0749Insightful D-branesBetter late than never: information retrieval from black holesThe Information paradox: A Pedagogical introductionarxiv-1009.0246New lower bounds for the border rank of matrix multiplication2012. Full version available at arXiv:1209.5993Explicit lower bounds via geometric complexity theoryarxiv-1801.10198arxiv-1803.05457arxiv-1905.07830arxiv-1909.05858arxiv-2309.08495arxiv-2410.22920arxiv-2602.15763paper-doi-10.1007-s00205-026-02198-0paper-doi-10.1017-jfm.2013.460web-4bd2b184cbeaweb-4fbe63501766web-52d3c45e6aa8math/preprints/The-Unique-Games-Theorem-September-23-2026/paper.pdf at main · openai/mathcordiverse/cordis: Meta-Framework of Spatiotemporal Composabilityweb-61d8f4540a88web-733f2ab74b7fweb-7d6e1a07349bweb-7ecf4c2a4e3dweb-a4bedf3302faweb-a7b433281f33web-f362ffba5311Stanford CME295 Transformers & LLMs | Autumn 2026 | Lecture 1 - TransformersStanford CME295 Transformers & LLMs | Autumn 2026 | Lecture 4 - Reinforcement Learning with LLMsThe Most Significant Moment In Math History Just HappenedStanford CME295 Transformers & LLMs | Autumn 2026 | Lecture 2 - Large Language ModelsState of AI 2026: The Biggest Stories of the Year (and AGI in 2027?), with Nathan BenaichNEW DeepSeek Harness - Super Agent!Stanford CME295 Transformers & LLMs | Autumn 2026 | Lecture 3 - LLM trainingETH FalseEfficient Estimation of Word Representations in Vector SpaceQuantum Computation vs. FirewallsAttention Is All You NeedRefusal in Language Models Is Mediated by a Single DirectionWait, Wait, Wait... Why Do Reasoning Models Loop?Giant Atlantic System will have big Impacts for Europe...GPT-6 Astra Made Qwen 6x Better at Math Without TrainingFrontierHarness EvalOpen Sourcing BERT: State-of-the-Art Pre-training for Natural Language ProcessinSparks of Weight-Space Programming: LLMs can natively read & write neural network weightsImproving language understanding with unsupervised learningGitHub - deepseek-ai/dsh-superagent-prototype: Towards a More Autonomous Agent SystemGitHub - deepseek-ai/deepseek-harness: DeepSeek Harness: Everything is a Plugin.∥u(t)∥L2(R3) < ∞, lim supP ?= NPzai-org/GLM-5 · Hugging FaceSteven Strogatz on the Humanity of Math#3-SATAcademic reward systemadversarial suffixAgent harnessAI adoptionAI-assisted mathematicsAI cybersecurityAI sovereigntyAMPS firewall paradoxApplied mathematicsArtificial general…Artificial intelligenceblack hole complementarityBlock-diagonal-plus-low-rank…Claude CodeCodexcoding agent setup with a…confidence buildup during…CordisData centerDeepSeekDeepSeek HarnessDeepSeek Sparse AttentionDeepSeek Super AgentDirectional ablationDoctoral educationdsh-superagentExponential Time Hypothesisfinite-time blowupFrontier AI modelFrontier Harness EvalGemma 3 270MGeometric complexity theoryGLM-4.7GLM-5GPT-6 Astragreedy decodingHawking radiation decodingHermite interpolationInference infrastructureknowledge distillationLow-rank matrixMATH-500Mathematical understandingMathematicsMathematics educationminimal human attention…Navier–Stokes equationsNavier–Stokes existence and…OpenAIP versus NP problemParity functionPermanentPlugin architecturePolynomial approximation of…proactive coding agentProximal policy optimizationquantum error correctionQwen3.5-0.8Brefusal directionrefusal in large language…Repetitive reasoningrisk aversion from hardness…Scientific authorshipself-similar collapsing…slimeSteven StrogatzSycophancytemperature (sampling)temporally correlated errorsweight orthogonalizationWeight-space programmingScott Aaronson: The Mathocalypse…arxiv-0704.0229Black Holes as Mirrors: Quantum…arxiv-0709.0749Insightful D-branesBetter late than never…The Information paradox: A…arxiv-1009.0246New lower bounds for the border…2012. Full version available at…Explicit lower bounds via…arxiv-1801.10198arxiv-1803.05457arxiv-1905.07830arxiv-1909.05858arxiv-2309.08495arxiv-2410.22920arxiv-2602.15763paper-doi-10.1007-s00205-026-0219…paper-doi-10.1017-jfm.2013.460web-4bd2b184cbeaweb-4fbe63501766web-52d3c45e6aa8math/preprints/The-Unique-Games-T…cordiverse/cordis: Meta-Framework…web-61d8f4540a88web-733f2ab74b7fweb-7d6e1a07349bweb-7ecf4c2a4e3dweb-a4bedf3302faweb-a7b433281f33web-f362ffba5311Lecture 1: TransformersLecture 4: Reinforcement learning with…Scott Aaronson on AI overtaking humans…Lecture 2: Large language modelsState of AI 2026 with Nathan Benaich…DeepSeek Super Agent, explained…Lecture 3: Training LLMsIs the Exponential Time Hypothesis…Efficient Estimation of Word…Quantum computation vs. black-hole…Attention Is All You NeedRefusal in Language Models Is…Wait, Wait, Wait... Why Do…Brian Shields forecasts rain and wind…Where GPT-6 Astra's math gain on Qwen…FrontierHarness Eval: nine agent…Open sourcing BERT (Google, 2018)GPT-6 Astra edits another network's…Improving language understanding…DeepSeek's dsh-superagent: a research…DeepSeek Harness: an open agent…A claimed finite-time blowup for the 3D…P vs NP, explained: Scott Aaronson's…GLM-5: zai-org's open model for…Steven Strogatz and Grant Sanderson on…