Tag
#interpretability
Reports tagged interpretability.
Reports
2 reportsOct 2026
10-25 inside-the-model How Language Models Use Ideas Before Putting Them Into Words To understand how a model can use an intermediate result without writing it down, follow what it represents, how later computation uses it, and what changes when it becomes language. 10-24 inside-the-model How a Model’s Numbers Become Words An activation can be translated into a sentence by giving language a computational job. Understanding that job reveals what the sentence preserves—and what still needs testing before it explains the model’s behavior.