Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

MemoChat: Tuning LLMs to Use Memos for Consistent Long-Range Open-Domain Conversation

About

We propose MemoChat, a pipeline for refining instructions that enables large language models (LLMs) to effectively employ self-composed memos for maintaining consistent long-range open-domain conversations. We demonstrate a long-range open-domain conversation through iterative "memorization-retrieval-response" cycles. This requires us to carefully design tailored tuning instructions for each distinct stage. The instructions are reconstructed from a collection of public datasets to teach the LLMs to memorize and retrieve past dialogues with structured memos, leading to enhanced consistency when participating in future conversations. We invite experts to manually annotate a test set designed to evaluate the consistency of long-range conversations questions. Experiments on three testing scenarios involving both open-source and API-accessible chatbots at scale verify the efficacy of MemoChat, which outperforms strong baselines. Our codes, data and models are available here: https://github.com/LuJunru/MemoChat.

Junru Lu, Siyu An, Mingbao Lin, Gabriele Pergola, Yulan He, Di Yin, Xing Sun, Yunsheng Wu• 2023

Related benchmarks

TaskDatasetResultRank
Long-form DialogueMT-Bench+
Quality Score85.83
32
Long-form DialogueSCM4LLMs
Quality83.47
32
Long-form DialogueLocomo
EM9.69
32
Multi-turn conversationLong-MT-Bench+
Accuracy1.88
10
Goal DetectionATOD Medium (test)
Goal Detection F180.38
9
Memory DiscriminationBID-20K
Acc65.6
9
Multi-turn conversationMT-Eval
Accuracy6.4
9
Goal DetectionATOD Complex (test)
Goal Detection F158.07
9
Memory DiscriminationIID-10K
Accuracy0.163
9
Status TrackingATOD Medium (test)
Status Tracking Accuracy73.88
9
Showing 10 of 14 rows

Other info

Follow for update