SyncAI.news, a Varaisys broadcasting
MemoryAthena: Adaptive Routing over Latent and Generated Memories
ML

Mingyuan Li, Guangsheng Yu, Juyuan Zhang, Xu Wang, Zhibo Man, Haonan Zhang, Shaoxiong Ji

· 1 min read

ResearcharXiv cs.CL

MemoryAthena: Adaptive Routing over Latent and Generated Memories

arXiv:2609.25853v1 Announce Type: new Abstract: Learned-memory methods store information in an explicit table and consume it through a separate reader, allowing addressing, storage, and reading to be modified independently. We study whether useful memory can also be generated rather than only retrieved. MemoryAthena uses three pathways: direct Engram retrieval (E), generation from retrieved Engram cues (GE), and generation from causal backbone states without consulting the memory table (GH). Generated memory is conditionally useful: it can complement E in one context but interfere with it in another. MemoryAthena therefore treats E as an anchor and learns when a generated representation should intervene. With the backbone, memory, generators, and readers frozen, a lightweight causal routing head is trained from counterfactual future-token likelihood advantages of GE and GH relative to E. At inference time, an admitted candidate modifies the E residual through bounded interpolation, while rejection recovers the direct pathway exactly. On question answering, MemoryAthena raises the five-task average from 37.65 to 39.28 over the direct pathway of the same checkpoint, while the six-task general-NLP average increases from 76.73 to 79.13. The complete memory-side system contains approximately 201M parameters, excluding the frozen backbone. Further analyses show complementary strengths among E, GE, and GH across tasks and inputs. These results support generated memory as a selective correction to direct retrieval and highlight routing when, which, and how strongly to intervene as the central challenge.

Original source

This story was published by arXiv cs.CL and written by Mingyuan Li, Guangsheng Yu, Juyuan Zhang, Xu Wang, Zhibo Man, Haonan Zhang, Shaoxiong Ji. SyncAI.news shows a preview; the complete article is on the publisher's site.

Read the full story on arxiv.org

Similar News