Search papers, labs, and topics across Lattice.
This paper introduces RILA, an execution-driven agent that enhances interactive web development by integrating browser rendering into the code generation process. By employing an Action Interaction Verification (AIV) module and an Execution-aware Rendering Score (ERS), RILA iteratively refines generated code based on real-time interaction feedback, ensuring both functional correctness and visual fidelity. The results demonstrate significant improvements in interaction and visual quality, with RILA achieving a notable performance boost in the compact Qwen3.5-9B model, surpassing larger models in the process.
RILA transforms web development by ensuring interactive functionality is prioritized alongside visual aesthetics, achieving superior performance with a compact model.
Multimodal large language models have achieved remarkable progress in front-end web development, generating interactive webpages from multimodal references such as screenshots and interaction videos. However, existing work largely emphasizes visual metrics such as aesthetics and layout similarity, while overlooking the more critical validation of interactive functionality. We present RILA, an execution-driven agent that puts browser rendering in the loop, iteratively editing generated code from runtime interaction feedback. RILA introduces an Action Interaction Verification (AIV) module that replays the reference interaction trajectory on the generated webpage to collect grounded execution-aware observations, and an Execution-aware Rendering Score (ERS) that jointly measures interaction correctness and visual fidelity to guide iterative optimization. We further build an execution-verified data synthesis pipeline that produces diverse, high-quality training data, offering gains complementary to inference-time optimization. On IWR-Bench, RILA consistently improves both interaction and visual fidelity across foundation models. Notably, with our training pipeline, RILA lifts the compact Qwen3.5-9B backbone from 40.40% to 57.52%, surpassing far larger one-shot generators, including the 1T-parameter Kimi-K2.6 (55.61%) and the proprietary GPT-5.5 (55.74%).
