Dynamic Long Short-Term Memory Based Memory Storage For Long Horizon LLM Interaction
This work addresses memory storage for LLMs to improve personalization in long interactions, but it is incremental as the LSTM-based memory encoder did not yield strong results.
The paper tackled the problem of enabling personalization in long conversations for Large Language Models (LLMs) by proposing Pref-LSTM, a framework that uses a BERT-based classifier to identify user preferences and a LSTM memory module for storage, with the classifier performing reliably in identifying explicit and implicit preferences.
Memory storage for Large Language models (LLMs) is becoming an increasingly active area of research, particularly for enabling personalization across long conversations. We propose Pref-LSTM, a dynamic and lightweight framework that combines a BERT-based classifier with a LSTM memory module that generates memory embedding which then is soft-prompt injected into a frozen LLM. We synthetically curate a dataset of preference and non-preference conversation turns to train our BERT-based classifier. Although our LSTM-based memory encoder did not yield strong results, we find that the BERT-based classifier performs reliably in identifying explicit and implicit user preferences. Our research demonstrates the viability of using preference filtering with LSTM gating principals as an efficient path towards scalable user preference modeling, without extensive overhead and fine-tuning.