Episodic Memory Benchmark / Episodic Memory Benchmark (short book)

Findings

IC-092All evaluated LLMs show consistent F1 degradation to at most 0.60 when two or more events match a retrieval cue [eval]
IC-093No evaluated LLM achieves perfect confabulation avoidance on questions about non-existent events [eval]
IC-094Episodic recall accuracy degrades systematically from content cues to space cues to time cues across all evaluated LLMs [eval]
IC-095Evaluated LLMs achieve at most 36% latest-state accuracy and 18% full-set accuracy on multi-event entity tracking, with low Kendall's tau on chronological ordering [eval]