English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Memory-augmented Large Language Models (LLMs) have demonstrated remarkableperformance in long-term human-machine interactions, which basically relies oniterative recalling and reasoning of history to generate high-qualityresponses. However, such repeated recall-reason steps easily produce biasedthoughts, \textit{i.e.}, inconsistent reasoning results when recalling the samehistory for different questions. On the contrary, humans can keep thoughts inthe memory and recall them without repeated reasoning. Motivated by this humancapability, we propose a novel memory mechanism called TiM (Think-in-Memory)that enables LLMs to maintain an evolved memory for storing historical thoughtsalong the conversation stream. The TiM framework consists of two crucialstages: (1) before generating a response, a LLM agent recalls relevant thoughtsfrom memory, and (2) after generating a response, the LLM agent post-thinks andincorporates both historical and new thoughts to update the memory. Thus, TiMcan eliminate the issue of repeated reasoning by saving the post-thinkingthoughts as the history. Besides, we formulate the basic principles to organizethe thoughts in memory based on the well-established operations,(\textit{i.e.}, insert, forget, and merge operations), allowing for dynamicupdates and evolution of the thoughts. Furthermore, we introduceLocality-Sensitive Hashing into TiM to achieve efficient retrieval for thelong-term conversations. We conduct qualitative and quantitative experiments onreal-world and simulated dialogues covering a wide range of topics,demonstrating that equipping existing LLMs with TiM significantly enhancestheir performance in generating responses for long-term interactions.

Think-in-Memory: Recalling and Post-thinking Enable LLMs with Long-Term Memory