From Emotion Classification to Emotional Reasoning: Enhancing Emotional Intelligence in Large Language Models

TL;DR

This paper explores how synthetic emotional reasoning data can enhance the emotional intelligence of small to medium-sized language models through fine-tuning, without changing their architecture.

Contribution

It introduces a multi-agent pipeline to generate therapy-style conversations and structured emotion MCQs, demonstrating improved emotional reasoning in 7B models.

Findings

Mistral 7B's EU score increased from 10.5 to 20.5

Mistral 7B's EA score increased from 40.5 to 60.0

Synthetic data effectively enhances emotional understanding in LLMs.

Abstract

This work investigates whether synthetic emotional chain-of-thought data can improve the emotional reasoning abilities of smaller open large language models (LLMs). We design a multi-agent generation pipeline that produces therapy-style conversations and converts them into structured emotion multiple-choice questions (MCQs) with explanations. We propose that fine-tuning a variety of 7B models on this dataset should yield substantial gains in emotional understanding and emotional awareness on EmoBench-style evaluations, suggesting that emotional reasoning can be induced without architectural changes. Our results demonstrate that fine-tuned Mistral 7B achieves EU improvements from 10.5 to 20.5 and EA improvements from 40.5 to 60.0, validating the effectiveness of synthetic emotional reasoning data for enhancing model capabilities in nuanced emotional tasks.