Can Instruction Fine-Tuned Language Models Identify Social Bias through Prompting?
This work addresses the need for efficient bias measurement in language models, though it is incremental as part of an ongoing framework.
The paper tackles the problem of evaluating instruction fine-tuned language models' ability to identify social biases through zero-shot prompting, finding that Alpaca 7B achieves the best accuracy of 56.7% on this task.
As the breadth and depth of language model applications continue to expand rapidly, it is increasingly important to build efficient frameworks for measuring and mitigating the learned or inherited social biases of these models. In this paper, we present our work on evaluating instruction fine-tuned language models' ability to identify bias through zero-shot prompting, including Chain-of-Thought (CoT) prompts. Across LLaMA and its two instruction fine-tuned versions, Alpaca 7B performs best on the bias identification task with an accuracy of 56.7%. We also demonstrate that scaling up LLM size and data diversity could lead to further performance gain. This is a work-in-progress presenting the first component of our bias mitigation framework. We will keep updating this work as we get more results.