historie nádherně ukazuje evoluci těch equations 😄 např jak se šlo z klasický mechaniky do statistický mechaniky do klasický field theory do relativistický mechaniky to kvantový mechaniky a pak se to spojovalo přes relativistic quantum field theory (ale ještě nám chybí spojit nebo přeformulovat obecnou relativitu nebo standardní model) Stephen Wolfram, zbožňuju ho, miluju jeho matematickej koncept všech možných systémů 😄 [Ruliad -- from Wolfram MathWorld](https://mathworld.wolfram.com/Ruliad.html) "The ruliad may be defined as the entangled limit of everything that is computationally possible, i.e., the result of following all possible computational rules in all possible ways. The ruliad can be considered as the ultimate abstraction and generalization involving any aspects of the physical universe. In particular, while a computational system or mathematical theory requires certain choices to be made, there are no choices or outside inputs in a ruliad because it already contains everything." Stephen Wolfram o jeho novým modelu observerů co modelují svět tak jak ho modelují protože jsou takový typ observera v této Ruliádě a modelovali bychom svět jinak kdybychom byly jinačí typ observerů (v jiným podprostoru Ruliády) 😄 [Solving the Problem of Observers & ENTROPY | Stephen Wolfram - YouTube](https://www.youtube.com/watch?v=0YRlQQw0d-4) Unexpected emergent patterns are everywhere in all physical systems https://explorer.globe.engineer/?q=mathematical+physics , https://explorer.globe.engineer/?q=Mathematics, https://explorer.globe.engineer/?q=foundations+of+mathematics, https://explorer.globe.engineer/?q=applied+mathematics , https://explorer.globe.engineer/?q=theory+of+everything Teď je asi největší limitace u AI systémů nacpat víc complex systematic coherent reasoning, planning, generalizing, agentnost (autonomita), memory, factual groundedness, online learning, ovládatelnost, což mají celkem weak na složitější tasky když se škálují jejich capabilities, ale děláme v tom progress, buď přes composing LLMs v multiagent systémech, škálování, kvalitnější data a trénování, nebo poupravenou nebo překopanou architekturu, apod.... a nebo se ještě řeší fungující robotika: Survey of LLMs: [[2312.03863] Efficient Large Language Models: A Survey](<https://arxiv.org/abs/2312.03863>), [[2311.10215] Predictive Minds: LLMs As Atypical Active Inference Agents](<https://arxiv.org/abs/2311.10215>), [A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications](https://arxiv.org/abs/2402.07927) Reasoning: [Human-like systematic generalization through a meta-learning neural network | Nature](<https://www.nature.com/articles/s41586-023-06668-3>), [[2305.20050] Let's Verify Step by Step](<https://arxiv.org/abs/2305.20050>), [[2302.00923] Multimodal Chain-of-Thought Reasoning in Language Models](<https://arxiv.org/abs/2302.00923>), [[2310.09158] Learning To Teach Large Language Models Logical Reasoning](<https://arxiv.org/abs/2310.09158>), [[2303.09014] ART: Automatic multi-step reasoning and tool-use for large language models](<https://arxiv.org/abs/2303.09014>), [AlphaGeometry: An Olympiad-level AI system for geometry - Google DeepMind](<https://deepmind.google/discover/blog/alphageometry-an-olympiad-level-ai-system-for-geometry/>) Robotics: [Mobile ALOHA - A Smart Home Robot - Compilation of Autonomous Skills - YouTube](<[Mobile ALOHA - A Smart Home Robot - Compilation of Autonomous Skills - YouTube](https://www.youtube.com/watch?v=zMNumQ45pJ8>),) [Eureka! Extreme Robot Dexterity with LLMs | NVIDIA Research Paper - YouTube](<[Eureka! Extreme Robot Dexterity with LLMs | NVIDIA Research Paper - YouTube](https://youtu.be/sDFAWnrCqKc?si=LEhIqEIeHCuQ0W2p>),) [Shaping the future of advanced robotics - Google DeepMind](<https://deepmind.google/discover/blog/shaping-the-future-of-advanced-robotics/>), [Optimus - Gen 2 - YouTube](<[Optimus - Gen 2 | Tesla - YouTube](https://www.youtube.com/watch?v=cpraXaw7dyc>),) [Atlas Struts - YouTube](<https://www.youtube.com/shorts/SFKM-Rxiqzg>), [Figure Status Update - AI Trained Coffee Demo - YouTube](<[Figure Status Update - AI Trained Coffee Demo - YouTube](https://www.youtube.com/watch?v=Q5MKo7Idsok>),) [Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks - YouTube](<[Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks - YouTube](https://www.youtube.com/watch?v=Qob2k_ldLuw>)) Multiagent systems: [[2402.01680] Large Language Model based Multi-Agents: A Survey of Progress and Challenges](<https://arxiv.org/abs/2402.01680>) Pozměněný/alternativní architektury: [Mixture of experts - Wikipedia](<https://en.wikipedia.org/wiki/Mixture_of_experts>) (celkem nový a používá se dneska všude), [Mamba (deep learning architecture) - Wikipedia](<https://en.wikipedia.org/wiki/Mamba_(deep_learning_architecture)>), [[2305.13048] RWKV: Reinventing RNNs for the Transformer Era](<https://arxiv.org/abs/2305.13048>), [V-JEPA: The next step toward advanced machine intelligence](<https://ai.meta.com/blog/v-jepa-yann-lecun-ai-model-video-joint-embedding-predictive-architecture/>), [Active Inference](<https://mitpress.mit.edu/9780262045353/active-inference/>) Agency: [[2305.16291] Voyager: An Open-Ended Embodied Agent with Large Language Models](<https://arxiv.org/abs/2305.16291>), [[2309.07864] The Rise and Potential of Large Language Model Based Agents: A Survey](<https://arxiv.org/abs/2309.07864>), [Agents | Langchain](<https://python.langchain.com/docs/modules/agents/>), [GitHub - THUDM/AgentBench: A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR'24)](<https://github.com/THUDM/AgentBench>), [[2401.12917] Active Inference as a Model of Agency](<https://arxiv.org/abs/2401.12917>), [CAN AI THINK ON ITS OWN? - YouTube](<[The Free Energy Principle approach to Agency - YouTube](https://www.youtube.com/watch?v=zMDSMqtjays>),) [Artificial Curiosity Since 1990](<https://people.idsia.ch/~juergen/artificial-curiosity-since-1990.html>) Factual groundedness: [[2312.10997] Retrieval-Augmented Generation for Large Language Models: A Survey](<https://arxiv.org/abs/2312.10997>), [Perplexity](<https://www.perplexity.ai/>), [ChatGPT - Consensus](<https://chat.openai.com/g/g-bo0FiWLY7-consensus>) Memory: větší context window [Gemini 10 mil token context window](<https://twitter.com/mattshumer_/status/1759804492919275555>), nebo [vektorový databáze](<https://en.wikipedia.org/wiki/Vector_database>) Online learning: <https://en.wikipedia.org/wiki/Online_machine_learning> Planning: [[2402.01817] LLMs Can't Plan, But Can Help Planning in LLM-Modulo Frameworks](<https://arxiv.org/abs/2402.01817>), [[2401.11708v1] Mastering Text-to-Image Diffusion: Recaptioning, Planning, and Generating with Multimodal LLMs](<https://arxiv.org/abs/2401.11708v1>), [[2305.16151] Understanding the Capabilities of Large Language Models for Automated Planning](<https://arxiv.org/abs/2305.16151>) Generalizing: [[2402.10891] Instruction Diversity Drives Generalization To Unseen Tasks](<https://arxiv.org/abs/2402.10891>), [Automated discovery of algorithms from data | Nature Computational Science](<https://www.nature.com/articles/s43588-024-00593-9>), [[2402.09371] Transformers Can Achieve Length Generalization But Not Robustly](<https://arxiv.org/abs/2402.09371>), [[2310.16028] What Algorithms can Transformers Learn? A Study in Length Generalization](<https://arxiv.org/abs/2310.16028>), [[2307.04721] Large Language Models as General Pattern Machines](<https://arxiv.org/abs/2307.04721>), [A Tutorial on Domain Generalization | Proceedings of the Sixteenth ACM International Conference on Web Search and Data Mining](<https://dl.acm.org/doi/10.1145/3539597.3572722>), [[2311.06545] Understanding Generalization via Set Theory](<https://arxiv.org/abs/2311.06545>), [[2310.08661] Counting and Algorithmic Generalization with Transformers](<https://arxiv.org/abs/2310.08661>), [Neural Networks on the Brink of Universal Prediction with DeepMind’s Cutting-Edge Approach | Synced](<https://syncedreview.com/2024/01/31/neural-networks-on-the-brink-of-universal-prediction-with-deepminds-cutting-edge-approach/>), [[2401.14953] Learning Universal Predictors](<https://arxiv.org/abs/2401.14953>), [Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks | Nature Communications](<https://www.nature.com/articles/s41467-021-23103-1>) Ovládatelnost řeší celej mechanistic interpretability obor [GitHub - JShollaj/awesome-llm-interpretability: A curated list of Large Language Model (LLM) Interpretability resources.](<https://github.com/JShollaj/awesome-llm-interpretability>), [Statistical learning theory](<https://en.wikipedia.org/wiki/Statistical_learning_theory>) a deep learning theory obor, [[2106.10165] The Principles of Deep Learning Theory](<https://arxiv.org/abs/2106.10165>), nebo jiný alignment a empirický alchemistický metody [[2309.15025] Large Language Model Alignment: A Survey](<https://arxiv.org/abs/2309.15025>) A je dost možný (a velký % researchers si myslí) že research snažící se ovládat tyhle šílený inscrutabe matrices nemá dostatečně rychlej development v porovnávání s capabilities researchem, např nemáme tušení jak vypnout behaviors s dosavadníma metodama [Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training \ Anthropic](<https://www.anthropic.com/news/sleeper-agents-training-deceptive-llms-that-persist-through-safety-training>), což teď poslední dny šlo vidět s tím jak GPT4 začlo outputovat totální chaos po updatu [OpenAI’s ChatGPT Went Completely Off the Rails for Hours](<https://www.thedailybeast.com/openais-chatgpt-went-completely-off-the-rails-for-hours>), Gemini bylo víc woke než byo intended, nebo každou chvíl vidím novej jailbreak co obchází zábrany :PepeLaugh~1: [[2307.15043] Universal and Transferable Adversarial Attacks on Aligned Language Models](<https://arxiv.org/abs/2307.15043>). AGI definitions: [[2311.02462] Levels of AGI for Operationalizing Progress on the Path to AGI](https://arxiv.org/abs/2311.02462) https://twitter.com/IntuitMachine/status/1721845203030470956 [[0712.3329] Universal Intelligence: A Definition of Machine Intelligence](https://arxiv.org/abs/0712.3329) Joscha Bach [LLMs and Reasoning - @lexfridman - YouTube](https://www.youtube.com/watch?v=F8-HEkm-SRc) [What is a hypergraph in Wolfram Physics?](https://lasttheory.com/article/what-is-a-hypergraph-in-wolfram-physics) [Mamba Might Just Make LLMs 1000x Cheaper... - YouTube](https://www.youtube.com/watch?v=SbmETE7Ey20) Flying drone wheel robot https://twitter.com/khademinori/status/1761844043548352550?t=82_s9tew6aFTsf7fDHkyAg&s=19 https://twitter.com/xiao_ted/status/1761865996716114412 long token context windows