Papers with memory capacity
Free Lunch for Efficient Textual Commonsense Integration in Language Models (2023.acl-long)
Copied to clipboard
| Challenge: | Recent years have witnessed the emergence of textual commonsense knowledge bases, aimed at providing more nuanced and context-rich knowledge. |
| Approach: | They propose to group training samples with similar commonsense descriptions into a single batch and reuse the encoded description across multiple samples. |
| Outcome: | The proposed method reduces the computational cost while preserving performance on larger datasets and on devices with more memory capacity. |
OCR-Memory: Optical Context Retrieval for Long-Horizon Agent Memory (2026.acl-long)
Copied to clipboard
| Challenge: | Existing LLMs are limited by text-context budgets, resulting in token-expensive storage of raw trajectories . Optical Context Retrieval Memory (OCR-Memory) renders historical tra-jectorios into images annotated with unique visual identifiers. |
| Approach: | They propose a framework that leverages the visual modality as a high-density representation of agent experience. |
| Outcome: | Optical Context Retrieval Memory (OCRM) renders historical trajectories into images annotated with unique visual identifiers. |
StructKV: Preserving the Structural Skeleton for Scalable Long-Context Inference (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing compression approaches prioritize tokens based on local saliency metrics to decouple prefill computation from decoding memory. |
| Approach: | They propose a structure-aware KV cache compression framework that prioritizes tokens based on local saliency metrics to decouple prefill computation from decoding memory. |
| Outcome: | The proposed framework preserves long-range dependencies and retrieval robustness. |
Knowledge Decoupling via Orthogonal Projection for Lifelong Editing of Large Language Models (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for enhancing large language models (LLMs) have achieved some success, but their knowledge understanding and memory capacity significantly degrades after extensive editing. |
| Approach: | They propose a method that stores the basis vectors of the representation space of past edits in a knowledge cache and projects the gradient of the current edit onto a space orthogonal to previous knowledge for updating. |
| Outcome: | The proposed method improves question-answering ability and hallucination mitigation by 14% and 61% for large language models after 3,000 edits. |
Deconstructing word embedding algorithms (2020.emnlp-main)
Copied to clipboard
| Challenge: | Word embeddings are reliable feature representations of words used in many NLP tasks today. |
| Approach: | They propose to deconstruct Word2vec, GloVe and others into a common form . they propose to generalize several word embedding algorithms into . a low rank embedder framework is proposed to generalise the algorithms into one common form. |
| Outcome: | The proposed framework can be used to make word embeddings more performant. |