We present an approach to managing short-term memory in chatbots. We use a combination of storage and automatic summarization techniques to optimize conversational context. This approach not only improves the fluidity of interactions but also ensures contextual continuity during long dialogue sessions. The code uses PyTorch and Hugging Face's transformers to manage and compress the conversation history.

Introduction

There's one problem when implementing your own chatbot, and that's memory management during the conversation. Of course, you can use ready-made libraries such as 'Langchain', 'Ollama', etc ... But what if you want to implement your own algorithm from scratch?

We present here an approach to managing short-term memory in chatbots, using a combination of storage and automatic summarization techniques to optimize conversational context. The introduced method relies on a dynamic memory structure that limits data size while preserving essential ...