The Curious Engineer

All blogs / AI / ML

Gonzo ML

gonzoml.substack.com

On modern ML in simple words. Click to read Gonzo ML, by Grigory Sapunov, a Substack publication with thousands of subscribers.

AI / ML

’23’24’25’26
Posts
118 from 23 Mar 2023 to 10 May 2026
Posting
monthly (15 in the last year)
Kind
personal blog
Feed
https://gonzoml.substack.com/feed

2026 11

  1. Recursion strikes back
  2. Growing Neural Cellular Automata
  3. Why I Keep Coming Back to Universal Transformers
  4. I read 47,094 ClawHub skills so your agent doesn’t have to
  5. How We Built a Full Browser Game in Two Evenings with Claude
  6. Do Code Agents Actually Understand the Code They're Working With?
  7. Deep Learning Legends: Ilya's List, or 90% of Everything That Matters in AI
  8. What is Life?
  9. 2026+
  10. 2025: The Year in Review
  11. Universal Reasoning Model

2025 38

  1. Self-Replicating Programs: Life's Origin Story Told Through Brainfuck and Assembly
  2. NeurIPS 2025 Best Papers in Comics
  3. The Transformer Zoo Revisited
  4. Visualizing Research: How I Use Gemini 3.0 to Turn Papers into Comics
  5. Tiny Recursive Model
  6. Hierarchical Reasoning Model
  7. Stochastic activations
  8. On people and neutrino
  9. V-JEPA 2: Scaling V-JEPA
  10. Tversky Neural Networks
  11. The ICML 2025 Outstanding Papers
  12. DolphinGemma
  13. Pay Attention to Small Weights
  14. The ArXiv Problem
  15. Darwin Gödel Machine
  16. Are Deeper LLMs Smarter, or Just Longer?
  17. AlphaEvolve
  18. The Next AI Frontier
  19. Muon Optimizer Accelerates Grokking
  20. ThoughtTerminator
  21. Chain Of Continuous Thought (Coconut)
  22. Intuitive Physics Emergence in V-JEPA
  23. Hunyuan-T1 & TurboS
  24. Leaked: OpenAI’s Upcoming Open Model
  25. Normalization-Free Transformers
  26. Generative Agents 2.0
  27. Analyze research papers with Gemini 2.0
  28. s1: Simple test-time scaling
  29. DeepSeek-R1: Open model with Reasoning
  30. DeepSeek-V3: Training
  31. DeepSeek-V3: Technical Details
  32. Not All Layers Are Equal
  33. 🤘ACDC (not that one)
  34. DeepSeek moment
  35. Transformer^2: Self-adaptive LLMs
  36. NVIDIA's Vision for AI
  37. LCM: Large Concept Model
  38. Quick reflection on AI in 2024

2024 43

  1. BLT: Byte Latent Transformer
  2. ModernBERT, the BERT of 2024
  3. Star Attention: Efficient LLM Inference over Long Sequences
  4. The Super Weight in Large Language Models
  5. JAX things to watch for in 2025
  6. Diffusion Models are Evolutionary Algorithms
  7. Make softmax great again
  8. Deep Learning Frameworks
  9. Listen to your papers
  10. Discovering Shetland(ic)
  11. Gödel Agent
  12. The Fall of the Digital Babel
  13. “Deep Learning with JAX” is out!
  14. Were RNNs All We Needed?
  15. Jamba-1.5: Hybrid Transformer-Mamba Models at Scale
  16. Transformer Layers as Painters
  17. LayerShuffle
  18. Chain-of-Thought → Whiteboard-of-Thought
  19. TextGrad: Automatic "Differentiation" via Text
  20. Superconducting Supercomputers
  21. (Perhaps) Beyond Human Translation: Harnessing Multi-Agent Collaboration for Translating Ultra-Long Literary Texts
  22. Open-Endedness is Essential for Artificial Superhuman Intelligence
  23. Mamba-2 is here!
  24. You Only Cache Once: Decoder-Decoder Architectures for Language Models
  25. Unconventional Values
  26. Optimizing large-context models
  27. xLSTM: Extended Long Short-Term Memory
  28. Yes, we KAN!
  29. Dejavu Transformers
  30. Chronos: Learning the Language of Time Series
  31. Many-Shot In-Context Learning
  32. Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Models
  33. [DeepMind SIMA] Scaling Instructable Agents Across Many Simulated Worlds
  34. OLMo: Accelerating the Science of Language Models
  35. Big Post About Big Context
  36. Neural Network Diffusion
  37. More Agents Is All You Need
  38. Thermodynamic AI is getting hotter
  39. Optimizing Distributed Training on Frontier for Large Language Models
  40. Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws
  41. TinyLlama: An Open-Source Small Language Model
  42. GFlowNets
  43. “Human Compatible”, Stuart Russell

2023 26

  1. PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU
  2. For Distillation, Tokens Are Not All You Need
  3. Toward understanding the communication in sperm whales
  4. Mamba: Linear-Time Sequence Modeling with Selective State Spaces
  5. [S4] Efficiently Modeling Long Sequences with Structured State Spaces
  6. Conway’s Game of Life is Omniperiodic
  7. [Google] Gemini: A Family of Highly Capable Multimodal Models
  8. System 2 Attention (is something you might need too)
  9. 🪆Matryoshka Representation Learning
  10. Mindstorms in Natural Language-Based Societies of Mind
  11. The convolution empire strikes back
  12. Sparse Universal Transformer
  13. MemWalker
  14. "Building Machines That Learn and Think Like People", 7 years later
  15. Chain-of-Thought → Tree-of-Thought
  16. Borges and AI
  17. GPT-4V is coming!
  18. Turing, “Intelligent Machinery, A Heretical Theory”, 1951
  19. Mortal Computers
  20. Large Language Model Programs
  21. Generative Agents: Interactive Simulacra of Human Behavior
  22. Uncovering mesa-optimization algorithms in Transformers
  23. Textbooks Are All You Need II: phi-1.5 technical report
  24. Learning to Model the World with Language
  25. The Guardian, 2024
  26. The Generative Cyberiad