February
Papers
- Small Batch Size Training for Language Models: When Vanilla SGD Works, and Why Gradient Accumulation Is Wasteful
- How can memories last for days, years, or a lifetime? Proposed mechanisms for maintaining synaptic potentiation and memory
Links
- https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16dd04 Welcome to Gas Town
- https://lucumr.pocoo.org/2026/1/18/agent-psychosis/ Agent Psychosis: Are We Going Insane?
- https://www.pnas.org/doi/10.1073/pnas.2531115123 Quantifying the compressibility of the human brain
- https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/ Datacenters in space are a terrible, horrible, no good idea.
- https://alain-airom.medium.com/domain-specific-language-models-book-review-158a4f83bbfa Domain Specific Language Models Book Review
- https://blog.comma.ai/datacenter/ Owning a $5M data center
- https://trysansa.com/benchmark?dimension=censorship Sansa Bench - Sansa Bench is a benchmark specifically designed to test models on difficult, real-world tasks and use cases.
- https://alignment.anthropic.com/2026/hot-mess-of-ai/ The Hot Mess of AI: How Does Misalignment Scale with Model Intelligence and Task Complexity?
- https://www.nature.com/articles/s41467-021-27653-2 Brain-inspired global-local learning incorporated with neuromorphic computing
- https://blog.evjang.com/2021/02 Don't Mess with Backprop: Doubts about Biologically Plausible Deep Learning
- https://harshanu.space/en/tech/ccc-vs-gcc/ CCC vs GCC
- https://www.biorxiv.org/content/10.1101/2025.03.09.642245v1.abstract Illusions of Alignment Between Large Language Models and Brains Emerge From Fragile Methods and Overlooked Confounds
- https://keh0t0.github.io/EgoX/ EgoX: Egocentric Video Generation from a Single Exocentric Video
- https://www.isomorphiclabs.com/articles/the-isomorphic-labs-drug-design-engine-unlocks-a-new-frontier The Isomorphic Labs Drug Design Engine unlocks a new frontier beyond AlphaFold
- https://age-verifier.kibty.town/ discord/twitch/kick/snapchat age verifier
- https://blog.can.ac/2026/02/12/the-harness-problem/ I Improved 15 LLMs at Coding in One Afternoon. Only the Harness Changed.
- https://ifp.org/nlm/ Beyond AlphaFold: How to target investments to develop new AI modelks that can uncover natural laws
- https://geohot.github.io//blog/jekyll/update/2026/02/15/tiny-corp-product.html tiny corp's product -- a training box
- https://www.nature.com/articles/s41467-025-63771-x Efficient and robust temporal processing with neural oscillations modulated spiking neural networks
- https://arxiv.org/html/2512.23146v1 A Network of Biologically Inspired Rectified Spectral Units (ReSUs) Learns Hierarchical Features Without Error Backpropagation
- https://en.wikipedia.org/wiki/Scintillator Scintillator
- https://djdumpling.github.io/2026/01/31/frontier_training.html frontier model training methodologies
- https://www.sambent.com/i2p-2-11-0-ships-post-quantum-crypto-after-botnet-siege/ I2P 2.11.0 Ships Post-Quantum Crypto After Botnet Siege
- https://www.cell.com/cell-reports/fulltext/S2211-1247(26)00062-8 Goal-directed learning in cortical organoids
- http://adam.chlipala.net/frap/ Formal Reasoning About Programs
- https://arxiv.org/abs/2509.18480 SimpleFold: Folding Proteins is Simpler than You Think
- https://arxiv.org/abs/2310.07999 LEMON: Lossless model expansion
- https://arxiv.org/abs/1801.00062 Dendritic error backpropagation in deep cortical microcircuits
- https://corticallabs.com/research Cortical Labs - Our Research