Code, notebooks, and archived results for reproducing the analyses in “Context Sufficiency in Transformers” (Computers, 2026).
reproducible-research information-theory transformers pytorch representation-learning language-models markov-chains hidden-markov-models mechanistic-interpretability activation-patching causal-interventions context-sufficiency
-
Updated
Nov 1, 2026 - Jupyter Notebook