All AI memory papers

CareCall combines a memory management layer, LLM summarizer, and memory-augmented input over HyperCLOVA to store and reuse summaries of users’ Health, Meals, Sleep, Visited Places, and Pets across weekly calls. In deployment to 147 socially isolated adults, CareCall with long-term memory yielded higher Health-detail and Clinical-detail disclosure counts per call than CareCall without memory, and longer average call durations (87.89s vs 75.48s).

WISE introduces a dual parametric memory with Side Memory, Memory Routing, and Knowledge Sharding and Merging to separate pretrained and edited knowledge inside LLaMA, Mistral, and GPT-J. On ZsRE with LLaMA-2-7B at T = 100 edits, WISE reaches an average score of 0.90 versus 0.76 for MEMIT-MASS, while maintaining 1.00 locality.

LONGMEM augments a frozen GPT-2*-style backbone with a Residual SideNet, Cached Memory Bank, Memory Retrieval and Fusion, and Cross-Network Residual Connections to read and use long-term key–value memories. On ChapterBreak AO3, LONGMEM reaches 40.5% suffix identification accuracy with infinite in-memory context, compared to 28.3% for Memorizing Transformer under the same 1k in-context window.

Empowering Working Memory for Large Language Model Agents introduces a Working Memory Hub, Episodic Buffer, Interaction History Window, Central Processor, and External Environment Interface to give LLM agents persistent, structured working and episodic memory. Empowering Working Memory for Large Language Model Agents is a conceptual blueprint rather than a benchmarked system, so no quantitative MAIN_RESULT against specific baselines is reported.

HimNet combines a GNN Encoder, Node Memory Module, Graph Memory Module, and Graph Decoder to reconstruct graphs via stored normal patterns and score anomalies by reconstruction and approximation errors. On the DD biochemical dataset, HimNet achieves 80.6% AUC compared to 70.6% for PK-iF, a +10.0 point gain over this two-step baseline.

Hierarchical Neural Memory Network (HMNet) stacks multi-level latent memories z1–z3 with Event-write, Up-write, Down-write, Update, and Readout operations driven by Event Sparse Cross Attention. On DSEC-Semantic, HMNet-L3 reaches 57.4 mIoU with event–RGB fusion, improving over a ResNet-50 baseline at 54.1 mIoU while also reducing latency, and on GEN1 HMNet-B1 matches AED at similar mAP with 57% lower latency.

In-context Autoencoder (ICAE) adds a LoRA-adapted encoder, memory tokens and slots, and combined autoencoding plus language modeling pretraining on top of an unchanged Llama decoder to compress long contexts into short, reusable memory spans. On Pile autoencoding with Llama-7B, ICAE compresses 512-token contexts into 128 memory slots with 99.1% BLEU and 0.017 loss, and on the PWC dataset ICAE with 128 slots reaches a 74.2% win+tie rate against GPT-4 when built on Llama-2-7B-chat.

In search of dispersed memories connects associative memory networks, modern Hopfield networks, generative diffusion models, and the denoising loss to show that diffusion score networks encode Hopfield-like energy landscapes in their weights. In search of dispersed memories demonstrates that exact diffusion dynamics reach Pearson correlations of 0.995–0.996 with modern Hopfield iterations on denoising and completion tasks, while classical Hopfield networks reach only 0.700–0.741.

MaLP combines a Dual-Process enhanced Memory (DPeM), Working Memory, Short-Term Memory (STM), Long-Term Memory (LTM), a Coordinator C, and Retriever R to coordinate short- and long-term personalization around a PEFT-tuned LLM. On MaLP’s medical dialogue benchmark with LLaMA-7B, MaLP achieves 69.95% preference classification accuracy and a 91.53% win rate in response generation, improving ROUGE-L profile QA from 29.66 to 33.91 over the LoRA baseline.

MBPTrack combines a Decoupling Feature Propagation Module, BPLocNet, box-prior sampling, and point-to-reference aggregation to track 3D objects from point clouds using temporal memory and size-aware localization. On KITTI, MBPTrack achieves 70.3% Success and 87.9% Precision, improving over CXTrack’s 67.5%/85.3% by +2.8/+2.6.

MemoryBank combines Memory Storage, Memory Retrieval, and a Memory Updating Mechanism to maintain daily conversations, event summaries, and user portraits for long-term personalization. On a 10-day, 15-user simulated dialog benchmark with 194 probing questions, MemoryBank-powered SiliconFriend ChatGPT achieves 0.716 correctness and 0.912 contextual coherence, surpassing SiliconFriend ChatGLM and SiliconFriend BELLE.

Recurrent Neural Networks and Long Short-Term Memory Networks explains how Backpropagation Through Time, LSTM gates and cells, Gated Recurrent Units, bidirectional RNN, and ELMo fit into one dynamical-systems view. Recurrent Neural Networks and Long Short-Term Memory Networks mainly contributes a structured tutorial and survey rather than new benchmark numbers.

Think-in-Memory (TiM) augments an LLM agent with a hash-based Memory Cache, a Hash-based Mapping F(·), and Insert, Forget, Merge organization operations over inductive thoughts. On the Chinese part of the GVD dataset, TiM with ChatGLM raises contextual coherence from 0.428 to 0.665 compared to SiliconFriend.

PMA-Net augments Transformer self-attention with memory banks, prototype keys, and prototype values built via K-Means and k-NN over past activations. On COCO Karpathy test, PMA-Net reaches 131.5 CIDEr under cross-entropy training, a +3.7 gain over a re-trained Transformer baseline.

Working Memory Capacity of ChatGPT probes ChatGPT with verbal and spatial n-back blocks, adding noise, feedback, and chain-of-thought variants to stress working memory. Working Memory Capacity of ChatGPT finds d′ falls to around 1 at n=3 in most conditions and that GPT-4’s verbal n-back capacity far exceeds Bloomz, ChatGLM, and Vicuna baselines.

BayesPCN combines predictive coding, conjugate Bayesian updates over W 0:L, sequential importance sampling, and a diffusion-based forget mechanism to build a hierarchical associative memory that supports continual one-shot writes. On CIFAR10 and Tiny ImageNet hetero-associative tasks, BayesPCN matches offline GPCN with MSE as low as 0.0000 while online GPCN rises to 0.0791 MSE on CIFAR10 mask at sequence length 1024.

Episodic Memory Question Answering combines allocentric top-down semantic features, spatiotemporal memory, and a LingUNet-based question-answering model to localize answers on scene floorplans from egocentric RGB-D tours. On the EMQA benchmark built from Matterport3D tours, Episodic Memory Question Answering with temporal features achieves 29.11 IoU and 62.27 recall on top-down maps, improving over SMNetDecoder by 2.19 IoU and 18.41 recall.

Memory Maze combines an online reinforcement learning environment, an offline dataset, and an offline probing protocol to stress-test long-term memory using Dreamer, Dreamer (TBTT), IMPALA, and supervised probe networks. On Memory 9x9, Dreamer (TBTT) achieves a return of 33.2 compared to 23.4 for IMPALA, while humans reach 26.4 and the oracle reaches 34.8.

Incremental Prompting uses Span-based Event Detection, Episodic Memory Prompting, Experience Replay, and Knowledge Distillation to attach and grow type-specific prompt vectors across tasks. On MAVEN Task 5, Incremental Prompting achieves 50.12% micro F1 versus 39.34% for KT, a +10.78 point gain under class-incremental evaluation.

Knowledge Aware Finetuning (KAFT) augments QA data with relevant context, irrelevant context, counterfactual context, and empty context built from SQuAD 2.0, T-REx, QASC, and TriviaQA, and uses pretrained model’s answer to label irrelevant slices. KAFT achieves up to 24× higher controllability than Noisy Finetuning and up to 6× higher robustness than Noisy Finetuning on PaLM 540B while matching TriviaQA validation accuracy.

TRIME aligns token embeddings with in-batch contextualized representations using a contrastive training objective, combined with local memory, long-term memory, and external memory constructed via specialized batching strategies. On WIKITEXT-103, TRIME with external memory (TRIMELMext) reaches 15.37 test perplexity using a 247M Transformer, a 0.75 improvement over kNN-LM with continuous cache.

Universal Hopfield Networks decompose single-shot associative memory into similarity, separation, and projection, instantiating Hopfield networks, sparse distributed memories, dense associative memories, and modern continuous Hopfield networks within one energy-based framework. Universal Hopfield Networks then compare dot product, Euclidean, Manhattan, and other similarity functions, finding that Manhattan and Euclidean similarity often yield higher retrieval capacity and robustness than dot-product-based modern continuous Hopfield networks on MNIST, CIFAR10, and Tiny ImageNet.

Adaptive Retrieval combines parametric memory, non-parametric memories, and popularity-based thresholds to decide when to augment questions with BM25 or Contriever passages on POPQA and EntityQuestions. On POPQA, Adaptive Retrieval with GPT‑3 davinci‑003 plus GenRead and Contriever reaches 46.5% accuracy, 5.3 points above always-retrieving systems while cutting GPT‑3 API costs by about 50%.

Generalizable Episodic Memory (GEM) combines a parametric network Mθ, implicit memory-based planning, twin back-propagation process, and conservative estimation on single step to generalize episodic returns across trajectories. GEM achieves higher average returns than TD3, SAC, DDPG, and TD3+SIL on MuJoCo tasks like Ant-v2, HalfCheetah-v2, and Humanoid-v2 within 1M environment steps.