ERNIE-Image Technical Report

We introduce ERNIE-Image, an open-source text-to-image generation model built upon an 8B single-stream DiT architecture. ERNIE-Image aims to bridge the gap between current open-source models and leading closed-source systems through more effective mining of large-scale pre-training data and improved supervision quality throughout training. During pre-training, we adopt a bottom-up data construction pipeline that combines fine-grained image categorization, rich caption annotation, aesthetic assessment, and hierarchical sampling. This strategy reduces data noise while preserving long-tail concepts and detailed real-world knowledge, providing a stronger foundation for complex generation tasks. In the post-training stage, we use a top-down data construction pipeline for high-demand scenarios, diversify prompt annotations to better match real user inputs, and apply a stabilized DPO strategy to align the model with human aesthetic preferences. We further train ERNIE-Image-Turbo for efficient 8-NFE generation and propose MT-DMD to mitigate capability drift during distillation. To make the model easier to use in practical scenarios, we equip it with a lightweight Prompt Enhancer that expands concise user intents into structured visual descriptions. In addition, we develop ERNIE-Image-Aes, an industrial-grade aesthetic model, together with ERNIE-Image-Aes-1K, a human-annotated benchmark for realistic aesthetic evaluation. Extensive qualitative and quantitative experiments show that ERNIE-Image achieves leading performance among open-source models and approaches top-tier commercial models in instruction following, text rendering, and aesthetic quality. We release the trained models and aesthetic resources to facilitate further academic research and technical progress in the AIGC community.

Mitigating Conversational Inertia in Multi-Turn Agents

Large language models excel as few-shot learners when provided with appropriate demonstrations, yet this strength becomes problematic in multiturn agent scenarios, where LLMs erroneously mimic their own previous responses as few-shot examples. Through attention analysis, we identify conversational inertia, a phenomenon where models exhibit strong diagonal attention to previous responses, which is associated with imitation bias that constrains exploration. This reveals a tension when transforming few-shot LLMs into agents: longer context enriches environmental feedback for exploitation, yet also amplifies conversational inertia that undermines exploration. Our key insight is that for identical states, actions generated with longer contexts exhibit stronger inertia than those with shorter contexts, enabling construction of preference pairs without environment rewards. Based on this, we propose Context Preference Learning to calibrate model preferences to favor low-inertia responses over highinertia ones. We further provide context management strategies at inference time to balance exploration and exploitation. Experimental results across eight agentic environments and one deep research scenario validate that our framework reduces conversational inertia and achieves performance improvements.

TextEditBench: Evaluating Reasoning-aware Text Editing Beyond Rendering

Text rendering has recently emerged as one of the most challenging frontiers in visual generation, drawing significant attention from large-scale diffusion and multimodal models. However, text editing within images remains largely unexplored, as it requires generating legible characters while preserving semantic, geometric, and contextual coherence. To fill this gap, we introduce TextEditBench, a comprehensive evaluation benchmark that explicitly focuses on text-centric regions in images. Beyond basic pixel manipulations, our benchmark emphasizes reasoning-intensive editing scenarios that require models to understand physical plausibility, linguistic meaning, and cross-modal dependencies. We further propose a novel evaluation dimension, Semantic Expectation (SE), which measures reasoning ability of model to maintain semantic consistency, contextual coherence, and cross-modal alignment during text editing. Extensive experiments on state-of-the-art editing systems reveal that while current models can follow simple textual instructions, they still struggle with context-dependent reasoning, physical consistency, and layout-aware integration. By focusing evaluation on this long-overlooked yet fundamental capability, TextEditBench establishes a new testing ground for advancing text-guided image editing and reasoning in multimodal generation.

Laboratory evaluation of a wearable instrumented headband for rotational head kinematics measurement

Mild traumatic brain injuries (mTBI) are a highly prevalent condition with heterogeneous outcomes between individuals. A key factor governing brain tissue deformation and the risk of mTBI is the rotational kinematics of the head. Instrumented mouthguards are a widely accepted method for measuring rotational head motions, owing to their robust sensor-skull coupling. However, wearing mouthguards is not feasible in all situations, especially for long-term data collection. Therefore, alternative wearable devices are needed. In this study, we present an improved design and data processing scheme for an instrumented headband. Our instrumented headband utilizes an array of inertial measurement units (IMUs) and a new data-processing scheme based on continuous wavelet transforms to address sources of error in the IMU measurements. The headband performance was evaluated in the laboratory on an anthropomorphic test device, which was impacted with a soccer ball to replicate soccer heading. When comparing the measured peak rotational velocities (PRV) and peak rotational accelerations (PRA) between the reference sensors and the headband for impacts to the front of the head, the correlation coefficients (r) were 0.80 and 0.63, and the normalized root mean square error (NRMSE) values were 0.20 and 0.28, respectively. However, when considering all impact locations, r dropped to 0.42 and 0.34 and NRMSE increased to 0.5 and 0.41 for PRV and PRA, respectively. This new instrumented headband improves upon previous headband designs in reconstructing the rotational head kinematics resulting from frontal soccer ball impacts, providing a potential alternative to instrumented mouthguards.