Hindsight Memory-PRM: Supervising Memory Management with Auditable Hindsight Credit

• URL: http://arxiv.org/abs/2608.29605v1
• Date: Sun, 30 Aug 2026 06:53:13 GMT
• Title: Hindsight Memory-PRM: Supervising Memory Management with Auditable Hindsight Credit
• Abstract summary: Hindsight Memory-PRM exploits this audit trail twice.<n> offline to train an operation-conditioned memory-utility critic.<n>Online, where retrievals, citations, and one controlled deletion-and-reanswer per probe settle an intervention-calibrated entry-level presence credit.
• Score: 48.811961095386
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Memory operations of long-horizon LLM agents are hard to supervise: an operation's value is unobservable when it is taken. But they are special -- they leave machine-readable evidence in the trajectory: retrieval hits and answer-time citations. Hindsight Memory-PRM exploits this audit trail twice: offline to train an operation-conditioned memory-utility critic, and online, where retrievals, citations, and one controlled deletion-and-reanswer per probe settle an intervention-calibrated entry-level presence credit, propagated along version chains as an action-level proxy reward -- no per-operation human labels, no Monte-Carlo replay of continuations. On held-out LoCoMo a local 8B policy reaches 77.5% under a fixed shared reader, surpassing its API teacher (65.1%) and all reproduced external systems, at one eighth the context of Mem0's official operating point; on LongMemEval, 79.0%. Ablations attribute the gain to causal calibration rather than signal density, and the policy converges to a multi-version memory organization whose gains no tested open-loop baseline reproduces.

Related papers

• AGM: Achievement-Grounded Memory for Closed-Loop Agents with Frozen VLA Policies [17.09836507714724]
  We propose Achievement-Grounded Memory (AGM) for frozen vision-language-action (VLA) policies.<n>AGM represents a task as a subgoal sequence with a progress pointer and advances this memory only after the current subgoal is verified by physical evidence.<n>On the RoboMME Counting benchmark, AGM reaches on PickXTimes and on BinFill, surpassing the strongest memory-augmented baseline by points on average.
  arXiv  Detail & Related papers  (2026-08-30T03:50:49Z)
• MemArbiter: Decision-Time Memory Arbitration for Long-Horizon LLM Agents [14.82470944373943]
  We propose MemArbiter, a function-aware memory arbitration framework.<n>MemArbiter decomposes interaction histories into atomic items, organizes them into five functional Memory Banks.<n>MemArbiter achieves success rates of 82.8% and 92.5% under 500- and 750-token budgets.
  arXiv  Detail & Related papers  (2026-08-03T12:10:52Z)
• When Memory Becomes Authority: Benchmarking Authority Collapse at the Memory Consolidation Boundary [12.82709779460352]
  LLM agents adapt across tasks by consolidating heterogeneous interaction histories into reusable facts, preferences, observations, and rules.<n>We identify authority collapse, in which consolidation preserves a claim while erasing the source constraints governing its authorized use.<n>We show that memory-driven adaptation must preserve not only what was learned, but also the authority under which it may be reused.
  arXiv  Detail & Related papers  (2026-08-03T04:15:40Z)
• 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)
• Memory Depth, Not Memory Access: Selective Parametric Consolidation for Long-Running Language Agents [0.0]
  We study memory depth: durable goal-conditioned tendencies written into a small parametric store.<n>We introduce the loop-drift protocol, a controlled stress test in which the retrieval index remains intact while working context is unloaded.<n>We find that selective consolidation factorizes into two controllable dimensions: selection and actuation.
  arXiv  Detail & Related papers  (2026-06-25T09:42:56Z)
• Memory is Reconstructed, Not Retrieved: Graph Memory for LLM Agents [51.30250860677378]
  We propose MRAgent, a framework that combines an associative memory graph with an active reconstruction mechanism.<n>Running on a Cue-Tag-Content graph, our active reconstruction mechanism integrates reasoning directly into memory access.<n>Experiments on the LoCoMo benchmark and LongMemEval benchmark demonstrate significant improvements over strong baselines.
  arXiv  Detail & Related papers  (2026-06-04T11:29:46Z)
• WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction [72.1620416874118]
  Multimodal large language models are increasingly deployed as long-horizon agents.<n>Existing benchmarks measure recall over static dialogue, collapse memory into a single end-of-task accuracy, and reduce visual observations to captions.<n>We formulate multimodal agent memory as an Action-World Interaction Loop with an observable four-stage lifecycle.
  arXiv  Detail & Related papers  (2026-05-28T04:27:20Z)
• 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)
• MemMA: Coordinating the Memory Cycle through Multi-Agent Reasoning and In-Situ Self-Evolution [52.29303869050117]
  Memory-augmented LLM agents maintain external memory banks to support long-horizon interaction.<n>MemMA is a plug-and-play multi-agent framework that coordinates the memory cycle along both the forward and backward paths.
  arXiv  Detail & Related papers  (2026-03-19T10:15:59Z)

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.