How Agents Decide: Goodfire's Eric Bigelow on Critical Tokens, Phase Shifts, & In-Context Learning
Goodfire's Eric Bigelow explains how models decide, covering critical tokens, phase shifts and in-context learning, in a long interpretability deep dive.
More in Models
32:07ModelsRecursive Self-Improvement and the Future of AI
a16zrecursive self-improvementinterview
5:38ModelsDeepSeek Solved AI’s Memory Problem
DeepSeekexplainer
25:04ModelsDeepSeek-V4.1 Flash: The Most Insane Optimization So Far
DeepSeekopen weightsdeep dive





