FlashEval: Towards Fast and Accurate Evaluation of Text-to-image Diffusion Generative Models
This addresses the efficiency problem for researchers and developers in generative AI by enabling faster model evaluation, though it is incremental as it builds on subset selection methods.
The paper tackles the problem of computationally expensive evaluation of text-to-image diffusion models by proposing FlashEval, an iterative search algorithm to select a representative subset of text-image datasets for evaluation. The result is that a 50-item subset achieves comparable evaluation quality to a randomly sampled 500-item subset on COCO annotations, achieving a 10x evaluation speedup.
In recent years, there has been significant progress in the development of text-to-image generative models. Evaluating the quality of the generative models is one essential step in the development process. Unfortunately, the evaluation process could consume a significant amount of computational resources, making the required periodic evaluation of model performance (e.g., monitoring training progress) impractical. Therefore, we seek to improve the evaluation efficiency by selecting the representative subset of the text-image dataset. We systematically investigate the design choices, including the selection criteria (textural features or image-based metrics) and the selection granularity (prompt-level or set-level). We find that the insights from prior work on subset selection for training data do not generalize to this problem, and we propose FlashEval, an iterative search algorithm tailored to evaluation data selection. We demonstrate the effectiveness of FlashEval on ranking diffusion models with various configurations, including architectures, quantization levels, and sampler schedules on COCO and DiffusionDB datasets. Our searched 50-item subset could achieve comparable evaluation quality to the randomly sampled 500-item subset for COCO annotations on unseen models, achieving a 10x evaluation speedup. We release the condensed subset of these commonly used datasets to help facilitate diffusion algorithm design and evaluation, and open-source FlashEval as a tool for condensing future datasets, accessible at https://github.com/thu-nics/FlashEval.