#memory mechanisms

34 stories · 5 in 7 days · 15 in 30 days
Stories, products and papers tagged "memory mechanisms", newest first.
Nov 13
Chinese contentCoReMem:为连续编码任务设计代价内化的记忆框架

arXiv上这篇论文解决了编码智能体记忆该存什么、什么时候用的问题。它把读写记忆的成本算进训练目标,在ChainSWE基准上让27B模型解决率从21.0%提到25.7%,还少走17%的步数。做智能体记忆或SWE agent的团队值得对比看看。

arXiv cs.AI@Peiwen Li et al. (9)Source
Nov 10
Nov 7
Nov 6
PapersBuildersSelected22:47
Amazon research shows weight memory beats retrieval for agent recall

The practical implication is a hybrid memory design: keep a text store for recent and never-recorded facts, and distill stable user knowledge into model weights. Weight memory alone forgets older facts and cannot flag gaps, so the router matters. Engineers building personalised agents can test this against MemOS-style retrieval pipelines. The paper is available on arXiv (2611.05636) with a Chat with Paper interface from dair.ai.

elvis@omarsar0Source
Nov 5
PapersBuildersSelected00:51
Tsinghua paper shows AI memory deletion fails 98.5% of forget requests

Memory benchmarks so far test only storing and recalling facts, so unlearning failures have gone unmeasured; BEMULA gives that gap its first systematic yardstick. Teams building agents that retain user data should treat delete requests as an unverified UI state and engineer an explicit delete path with adversarial extraction tests. The paper is academic work with no regional availability angle, and its closest comparison is the memory features of mainstream agent frameworks rather than another benchmark.

rohanpaul_ai@rohanpaul_aiSource
Nov 3
PapersBuildersSelected02:06
Meta paper shows routed memory beats shared history for agent learning

Splitting agent memory by problem type, rather than one shared log, delivers both better task performance and much lower token spend at the same time. Practitioners building agents that handle mixed task types can apply this by routing each new problem to a memory pool of similar past attempts. The 65% to 72% API cost reduction makes it especially relevant for anyone running agents at volume on metered model APIs.

rohanpaul_ai@rohanpaul_aiSource
Oct 30
Oct 23
Oct 20
PapersBuildersSelected11:47
Addressed-Memory Writes Fail at Reasoning in Qwen3 and DeepSeek Beds

Memory write-injection looks like a cheap way to add knowledge to a deployed model, but these results show the added facts cannot be chained into multi-hop reasoning by the model itself — orchestration must live in the application layer. Builders wiring RAG-style memory into agents should chain written atoms explicitly in prompts or system logic, and benchmark vendors should adopt coverage, synergy and distortion reporting. The tested beds include Chinese frontier models Qwen3 and DeepSeek, making the findings broadly relevant across model families.

arXiv: DeepSeek@Wei Gao et al. (4)Source
Oct 16
Oct 9
PapersBuildersSelected08:58
Sakana AI's Continuous Memory Machine Adds Dual Memory to Recurrent Models

The dual-memory split addresses a core limitation of recurrent models, where short-term computation and long-term storage compete for a single hidden vector. Researchers working on state-tracking architectures or memory-augmented networks can compare CMM's results against CTM and Transformer baselines in the paper. Its selective use of long-term memory suggests a practical route to scaling recurrence beyond fixed context lengths.

DAIR.AI@dair_aiSource
Oct 8
Oct 1
Aug 26
Aug 18
Aug 10
Aug 4
Jul 28
Jul 17
Jun 30
Jun 19
Jun 17
Jun 15
Jun 13
Jun 9
Jun 4
Jun 2
May 17
May 13