MemTrace: Probing What Final Accuracy Misses in Long-Term Memory

• URL: http://arxiv.org/abs/2606.17328v1
• Date: Mon, 15 Jun 2026 22:21:23 GMT
• Title: MemTrace: Probing What Final Accuracy Misses in Long-Term Memory
• Abstract summary: We introduce MemTrace, a benchmark whose unit of measurement is the knowledge point.<n>MemTrace probes each fact along three controlled dimensions.<n>We find that similar pooled accuracy hides different failures.
• Score: 39.696350769003104
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: LLM agents increasingly maintain long-term memory of user facts across sessions. Yet such memory is usually evaluated by aggregating accuracy over question rows or episodes. Because this approach scores question rows independently, even when several questions probe the same fact, it cannot show how that fact behaves as conditions change. We introduce MemTrace, a benchmark whose unit of measurement is the knowledge point: a single typed fact about the user, rather than an individual question. MemTrace probes each fact along three controlled dimensions: memory age, defined by how many sessions ago the fact appeared in the history; question type, covering current state, earlier state, and trajectory of change; and evidence condition, covering present, missing, and contradicted-by-false-premise settings. Evaluating 13 memory-system configurations across four paradigms, we find that similar pooled accuracy hides different failures: recovering a fact's current and earlier states does not imply tracking how it changed, and safe abstention does not imply correcting a false premise. The dominant bottleneck is evidence use, not retrieval: when systems fail, the evidence was retrievable 10 times more often than it was missing. These results suggest that improving long-term memory requires better use of reachable evidence, not simply more storage or retrieval.

Related papers

• Procedural Memory Under Change: Reuse and Interference in Controlled Web Tasks [0.0]
  Procedural memory lets language agents reuse successful routines, but reuse presumes that a stored routine remains applicable.<n>We study what happens when that presumption is deliberately violated.
  arXiv  Detail & Related papers  (2026-09-09T06:22:42Z)
• MemOps: Benchmarking Lifecycle Memory Operations in Long-Horizon Conversations [50.24315431387575]
  We introduce MemOps, a benchmark that reformulates conversational memory as a sequence of lifecycle operations.<n>MemOps represents each memory event with a structured trace specifying its trigger, target, scope, state transition, and supporting evidence.<n>Across long-context, retrieval-based, parametric and managed-memory systems, MemOps disentangles failure modes that final-answer accuracy alone conceals.
  arXiv  Detail & Related papers  (2026-07-14T15:33:44Z)
• DynamicMem: A Long-Horizon Memory Benchmark in Real-World Settings [27.335953561941736]
  LLM agents increasingly act as personal assistants that must remember a user's profile over months.<n>Existing benchmarks evaluate this "memory" ability through short, simplified interactions.<n>We introduce DynamicMem, a benchmark that constructs 15 months of activity per user.
  arXiv  Detail & Related papers  (2026-06-22T05:41:20Z)
• RaMem: Contextual Reinstatement for Long-term Agentic Memory [21.228780482661076]
  We propose Contextual Reinstatement for Agentic Memory (RaMem), a framework that turns retrieved memory fragments into contextually evidence.<n>RaMem operates through four coordinated stages: (i) evidence anchoring grounds each memory in its original episodic conditions, especially event time, mention time, session span, and participants; (ii) recall condition induction derives the evidence conditions implied by the query; (iii) validity-aware retrieval uses these conditions to prioritize context-compatible memories while retaining content-relevant candidates as fallback evidence; and (iv) context-preserved synthesis keeps the selected memories' structured context available to the generator
  arXiv  Detail & Related papers  (2026-06-22T04:41:14Z)
• REAL: A Reasoning-Enhanced Graph Framework for Long-Term Memory Management of LLMs [17.526686616588794]
  Large Language Models (LLMs) are increasingly expected to interact with users over long time horizons.<n>LLMs cannot retain all past interactions, making long-term memory management essential for storing, updating, and retrieving historical information.<n> REAL constructs long-term conversational memory as a temporal and confidence-aware directed property graph.
  arXiv  Detail & Related papers  (2026-06-09T10:53:10Z)
• MemGuard: Preventing Memory Contamination in Long-Term Memory-Augmented Large Language Models [56.31411457917676]
  We introduce MemGuard, a type-aware memory framework that preserves functional memory boundaries during memory construction and retrieval.<n>Across hallucination and long-horizon conversation benchmarks, MemGuard improves memory reliability by up to 28.27% while retrieving up to 5.8x fewer memory tokens than prior methods.
  arXiv  Detail & Related papers  (2026-05-27T06:04:19Z)
• Causal Intervention-Based Memory Selection for Long-Horizon LLM Agents [0.33842793760651557]
  We propose a causal memory-selection technique that estimates how candidate memories affect the model's answer under controlled interventions.<n>Results show that CMI achieves a stronger balance between answer quality and robustness to misleading memory, suggesting that reliable long-term memory requires selecting context based on causal usefulness rather than relevance alone.
  arXiv  Detail & Related papers  (2026-05-17T20:21:55Z)
• Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM Memory [50.857546269660276]
  We introduce SeqMem-Eval, a diagnostic evaluation framework for sequentially evolving large language models (LLMs) memory.<n>Rather than focusing only on final performance, SeqMem-Eval evaluates how memory states evolve, generalize, consolidate experience, and retain useful information during sequential inference.
  arXiv  Detail & Related papers  (2026-05-14T20:15:22Z)
• MemEye: A Visual-Centric Evaluation Framework for Multimodal Agent Memory [80.97855900579512]
  Existing evaluations rarely test whether agents preserve the visual evidence needed for later reasoning.<n>We introduce MemEye, a framework that evaluates memory capabilities from two dimensions.<n>Under this framework, we construct a new benchmark across 8 life-scenario tasks.
  arXiv  Detail & Related papers  (2026-05-14T17:37:52Z)
• STALE: Can LLM Agents Know When Their Memories Are No Longer Valid? [5.361950931863979]
  Large Language Model (LLM) agents are increasingly expected to maintain coherent, long-term personalized memory.<n>Current benchmarks primarily measure static fact retrieval, overlooking the ability to revise stored beliefs when new evidence emerges.<n>We introduce STALE, a benchmark of 400 expert-validated conflict scenarios.<n>We present CUPMem, a prototype that strengthens write-time revision through structured state consolidation and propagation-aware search.
  arXiv  Detail & Related papers  (2026-05-07T16:31:15Z)
• Belief Memory: Agent Memory Under Partial Observability [56.41506249481312]
  We propose BeliefMem, which shifts the memory paradigm from committing to a single conclusion per observation to retaining multiple candidate conclusions with their probabilities.<n> BeliefMem preserves the uncertainty that the deterministic paradigm discards, enabling the agent to act with high confidence.<n> Empirical evaluations on LoCoMo and ALFWorld benchmarks show that, even with limited data, BeliefMem achieves the best average performance.
  arXiv  Detail & Related papers  (2026-05-07T02:03:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.