Call for Rigor in Reporting Quality of Instruction Tuning Data
This work highlights a methodological flaw in evaluating instruction tuning data quality, which is incremental but important for ensuring rigorous research practices in aligning large language models.
The paper identifies that arbitrary hyperparameter selection in instruction tuning studies can lead to unreliable conclusions about data quality, demonstrating this issue with experiments on LIMA and Alpaca datasets.
Instruction tuning is crucial for adapting large language models (LLMs) to align with user intentions. Numerous studies emphasize the significance of the quality of instruction tuning (IT) data, revealing a strong correlation between IT data quality and the alignment performance of LLMs. In these studies, the quality of IT data is typically assessed by evaluating the performance of LLMs trained with that data. However, we identified a prevalent issue in such practice: hyperparameters for training models are often selected arbitrarily without adequate justification. We observed significant variations in hyperparameters applied across different studies, even when training the same model with the same data. In this study, we demonstrate the potential problems arising from this practice and emphasize the need for careful consideration in verifying data quality. Through our experiments on the quality of LIMA data and a selected set of 1,000 Alpaca data points, we demonstrate that arbitrary hyperparameter decisions can make any arbitrary conclusion.