WISE

The 'WISE' dataset/benchmark contains a collection of image-text pairs and is used to evaluate the performance of multimodal models in tasks related to image generation and editing.

Papers using WISE (4)

Think-Then-Generate: Reasoning-Aware Text-to-Image Diffusion with LLM Encoders2026DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing2026Rethinking UMM Visual Generation: Masked Modeling for Efficient Image-Only Pre-training2026ARM: An AutoRegressive Large Multimodal Model with Unified Discrete Representations2026