T2I-ConBench: Text-to-Image Benchmark for Continual Post-training
This addresses a critical gap for researchers in continual learning and text-to-image generation by providing a comprehensive benchmark to accelerate progress, though it is incremental as it focuses on evaluation rather than new methods.
The paper tackles the lack of a standardized evaluation protocol for continual post-training in text-to-image models by introducing T2I-ConBench, a unified benchmark that assesses retention, performance, forgetting, and generalization across practical scenarios, finding that no existing method excels in all dimensions.
Continual post-training adapts a single text-to-image diffusion model to learn new tasks without incurring the cost of separate models, but naive post-training causes forgetting of pretrained knowledge and undermines zero-shot compositionality. We observe that the absence of a standardized evaluation protocol hampers related research for continual post-training. To address this, we introduce T2I-ConBench, a unified benchmark for continual post-training of text-to-image models. T2I-ConBench focuses on two practical scenarios, item customization and domain enhancement, and analyzes four dimensions: (1) retention of generality, (2) target-task performance, (3) catastrophic forgetting, and (4) cross-task generalization. It combines automated metrics, human-preference modeling, and vision-language QA for comprehensive assessment. We benchmark ten representative methods across three realistic task sequences and find that no approach excels on all fronts. Even joint "oracle" training does not succeed for every task, and cross-task generalization remains unsolved. We release all datasets, code, and evaluation tools to accelerate research in continual post-training for text-to-image models.