MemTrace shows that evidence utilization, not retrieval, is the dominant failure mode in LLM long-term memory systems across tested configurations.
hub
Memoryllm: Towards self-updatable large language models.arXiv preprint arXiv:2402.04624
15 Pith papers cite this work. Polarity classification is still indexing.
15
Pith papers citing it
abstract
Existing Large Language Models (LLMs) usually remain static after deployment, which might make it hard to inject new knowledge into the model. We aim to build models containing a considerable portion of self-updatable parameters, enabling the model to integrate new knowledge effectively and efficiently. To this end, we introduce MEMORYLLM, a model that comprises a transformer and a fixed-size memory pool within the latent space of the transformer. MEMORYLLM can self-update with text knowledge and memorize the knowledge injected earlier. Our evaluations demonstrate the ability of MEMORYLLM to effectively incorporate new knowledge, as evidenced by its performance on model editing benchmarks. Meanwhile, the model exhibits long-term information retention capacity, which is validated through our custom-designed evaluations and long-context benchmarks. MEMORYLLM also shows operational integrity without any sign of performance degradation even after nearly a million memory updates. Our code and model are open-sourced at https://github.com/wangyu-ustc/MemoryLLM.
hub tools
citation-role summary
background 2
method 1
citation-polarity summary
polarities
background 3representative citing papers
MemSearcher trains LLMs to manage compact memory in multi-turn searches via multi-context GRPO for end-to-end RL, outperforming ReAct-style baselines with stable token counts.
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
Explicit bank, retrieval, and QA state roles reduce ghost-memory failures on conflict-heavy LTP and improve some temporal scores on LoCoMo, with host-dependent gains.
RefMem-Bench benchmarks reflective memory in dialogue with 26K instances across eight dimensions, and REMIND improves model accuracy via hierarchical evidence retrieval, grounding, and abstraction.
FluxMem evolves memory as a heterogeneous graph via three refinement stages and reports consistent state-of-the-art results on LoCoMo, Mind2Web, and GAIA benchmarks.
Language models show superior memory to humans on psych experiments but can be adjusted via prompting and compaction to forget more human-like, yielding better user simulators.
In-Place TTT adapts LLM MLP projection matrices at test time with a next-token-aligned objective and chunk-wise updates, enabling better long-context performance as a drop-in enhancement.
HyMem introduces dual-granular memory storage with a lightweight summary module for fast responses and selective activation of a deep LLM module for complex queries, outperforming full-context baselines by 92.6% lower computational cost on LOCOMO and LongMemEval benchmarks.
MemoryAgentBench is a multi-turn benchmark covering four memory competencies, and current memory agents fail at selective forgetting and long-range understanding.
AgRefactor deploys a self-evolving multi-agent workflow that combines LLM rewrites with automated tools to convert software into HLS code, matching or beating baselines on long benchmarks and delivering 6.51x geometric mean speedup after optimization.
EMBER learns to retain budgeted, source-backed evidence capsules so long-horizon agents recover answer-relevant facts without rereading the full history.
The paper surveys human memory categories, maps them to LLM memory, and proposes a new three-dimension (object, form, time) categorization into eight quadrants to organize existing work and highlight open problems.
TrustMem introduces a verifier for memory update transitions and preference-guided RL to cut omission, corruption, and hallucination rates in LLM agent memory while reaching SOTA on MemoryAgentBench and HaluMem.
A survey of large-model inference optimization, organized as a four-layer 'token-operations' taxonomy: multi-model fusion, model optimization, compute-model fusion, and compute-network-model fusion.
citing papers explorer
-
MemTrace: Probing What Final Accuracy Misses in Long-Term Memory
MemTrace shows that evidence utilization, not retrieval, is the dominant failure mode in LLM long-term memory systems across tested configurations.
-
MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning
MemSearcher trains LLMs to manage compact memory in multi-turn searches via multi-context GRPO for end-to-end RL, outperforming ReAct-style baselines with stable token counts.
-
What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
-
A-TMA: Decoupling State-Aware Memory Failures in Long-Term Agent Memory
Explicit bank, retrieval, and QA state roles reduce ghost-memory failures on conflict-heavy LTP and improve some temporal scores on LoCoMo, with host-dependent gains.
-
Connecting the Dots: Benchmarking Reflective Memory in Long-Horizon Dialogue
RefMem-Bench benchmarks reflective memory in dialogue with 26K instances across eight dimensions, and REMIND improves model accuracy via hierarchical evidence retrieval, grounding, and abstraction.
-
Rethinking Memory as Continuously Evolving Connectivity
FluxMem evolves memory as a heterogeneous graph via three refinement stages and reports consistent state-of-the-art results on LoCoMo, Mind2Web, and GAIA benchmarks.
-
Simulating Human Memory with Language Models
Language models show superior memory to humans on psych experiments but can be adjusted via prompting and compaction to forget more human-like, yielding better user simulators.
-
In-Place Test-Time Training
In-Place TTT adapts LLM MLP projection matrices at test time with a next-token-aligned objective and chunk-wise updates, enabling better long-context performance as a drop-in enhancement.
-
HyMem: Hybrid Memory Architecture with Dynamic Retrieval Scheduling
HyMem introduces dual-granular memory storage with a lightweight summary module for fast responses and selective activation of a deep LLM module for complex queries, outperforming full-context baselines by 92.6% lower computational cost on LOCOMO and LongMemEval benchmarks.
-
Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions
MemoryAgentBench is a multi-turn benchmark covering four memory competencies, and current memory agents fail at selective forgetting and long-range understanding.
-
AgRefactor: Self-Evolving Agentic Workflow for HLS Compatibility and Performance
AgRefactor deploys a self-evolving multi-agent workflow that combines LLM rewrites with automated tools to convert software into HLS code, matching or beating baselines on long benchmarks and delivering 6.51x geometric mean speedup after optimization.
-
EMBER: Efficient Memory via Budgeted Evidence Retention for Long-Horizon Agents
EMBER learns to retain budgeted, source-backed evidence capsules so long-horizon agents recover answer-relevant facts without rereading the full history.
-
From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs
The paper surveys human memory categories, maps them to LLM memory, and proposes a new three-dimension (object, form, time) categorization into eight quadrants to organize existing work and highlight open problems.
-
TRUSTMEM: Learning Trustworthy Memory Consolidation for LLM Agents with Long-Term Memory
TrustMem introduces a verifier for memory update transitions and preference-guided RL to cut omission, corruption, and hallucination rates in LLM agent memory while reaching SOTA on MemoryAgentBench and HaluMem.
-
Token-Operations-Oriented Inference Optimization Techniques for Large Models
A survey of large-model inference optimization, organized as a four-layer 'token-operations' taxonomy: multi-model fusion, model optimization, compute-model fusion, and compute-network-model fusion.