When Should LLM Agents Forget? Memory Filtering Under Query-Critical Resolver Errors
The work examines when language-model agents should discard stored information, focusing on memory filtering and resolver errors that matter to later queries.
ArXiv CS.AI
Source
LLM memory systems often delete facts classified as stale. We show that this hard-filtering step can reduce downstream accuracy when resolver errors delete query-critical facts. Across three models, a controlled sensitivity intervention moves hard-drop from gains of +0.186 to +0.230 at perfect resolution to losses of -0.176 to -0.181 at the lowest tested reliability. Paired targeted interventions show that false-drop causes significantly greater degradation than false-keep, while a non-deleting baseline can still fail when incorrect labels steer generation. On a frozen 102/102 calibration/test split, a simple deployment-matched gate reaches 0.655 accuracy versus 0.641 for always hard-drop and 0.662 for test-informed oracle selection. Global-accuracy crossovers require gold labels and are model- and task-specific diagnostics, not deployment thresholds. Deletion should instead be treated as cost-sensitive: validate query-critical false-drop risk on matched data, or default to label-free retention when such validation is unavailable.
Generated insight by OnAirToday AI
An agent’s retained information can influence how it responds to later queries. Studying memory filtering alongside query-critical resolver errors highlights the importance of deciding what to keep or discard.

Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.