Adam Scherlis

Affiliation

• Research at d_model (dmodel.ai) [2025 –]

Blog posts (Research)

• (dmodel) Discovering Concept-Editing Algorithms With LLM Agents [2026-07-01]
• (LW) Estimating the Probability of Sampling a Trained Neural Network at Random [2025-02-28]
• (EleutherAI) Experiments in Weak-to-Strong Generalization [2024-06-14]

Blog posts (Fun)

• Utilitarian inequality metrics [2025-11-11]
• Ternary plots are underrated [2025-11-10]
• The grapefruit juice effect [2025-11-09]
• n-ary Huffman coding [2025-11-08]
• Review: Parsifal at the SF Opera [2025-11-07]
• GPTF-8: A tokenizer-based character encoding [2025-11-06]
• Review: K-Pop Demon Hunters (2025) [2025-11-05]
• Maxwell's Demon and the Arrow of Time [2025-11-04]
• Rainbows, fractals, and crumpled paper: Hölder continuity [2025-11-03]
• FTL travel and scientific realism [2025-11-01]
• A factorial puzzle [2025-02-25]
• Hex runes [2025-01-07]
• Randomly-generated shape pendants [2025-01-05]
• Morse code prosigns [2025-01-04]
• Iterated log coding [2025-01-01]
• Math facts: 2025 [2024-12-31]
• New Blog! [2024-12-30]

Other blogs

• LessWrong [2021 –]
• WordPress [2010 – 2024]

Papers

• arXiv
• Google Scholar

Code

• GitHub

Contact

• Twitter
• Email