FronTalk: New Benchmark for Front-End Code Generation with Visual and Conversational Feedback
Researchers introduced FronTalk, a benchmark dataset of 100 multi-turn dialogues for evaluating front-end code generation systems that use both text and visual instructions. The benchmark revealed two major challenges: models frequently overwrite previously implemented features (forgetting issue) and struggle to interpret visual feedback, particularly open-source vision-language models. This work addresses a gap in understanding how AI systems handle multi-modal, conversational code generation in realistic front-end development scenarios.
FronTalk is a new benchmark for evaluating front-end code generation systems in conversational, multi-modal settings. The dataset comprises 100 multi-turn dialogues derived from real-world websites across domains including news, finance, and art, with each interaction turn containing both textual and visual instructions representing the same user intent. The researchers evaluated 20 different models using a novel agent-based evaluation framework that simulates user interactions and measures both functional correctness and user experience. Key findings identified a significant forgetting issue where models overwrite previously implemented features during multi-turn interactions, and persistent difficulties in interpreting visual feedback, especially for open-source vision-language models. To address the forgetting problem, the authors proposed AceCoder, a baseline method using autonomous web agents to critique each implementation step, which reduced forgetting nearly to zero and improved performance by up to 9.3 percentage points (from 56.0% to 65.3%). The benchmark and code are publicly released to support future research in multi-turn, multi-modal code generation.
What different sources said
- arXiv cs.LGCenter
FronTalk: Benchmarking Front-End Development as Conversational Code Generation with Multi-Modal Feedback
Related
Genetic Drift, Not Selection, Drives Rapid Feather Color Evolution in Island Bird Radiation
A new study of an island bird radiation found that rapid evolution of feather coloration is driven primarily by genetic drift in small populations rather than sexual or ecological selection. The research integrated whole-genome data with detailed plumage measurements across complete species sampling to test whether signaling trait evolution correlates with speciation rates. The findings suggest that neutral demographic processes play a central role in generating phenotypic diversity during island radiations, challenging assumptions about the mechanisms driving rapid evolution.
New AI Model Improves Prediction of Therapeutic Peptide Function from Protein Sequences
Researchers developed a lightweight CNN classifier that predicts whether peptide sequences have therapeutic properties, trained on a database of 54,655 peptides across 48 functional categories. The model uses a novel negative sampling strategy to reduce false positive rates from over 60% in previous approaches to 2.1%. This advancement could accelerate drug discovery by enabling faster computational screening of peptide candidates before expensive experimental testing.
Study Shows Different Metabolic Stress Models Produce Distinct Effects on Human Neuronal Networks
Researchers tested three common in vitro metabolic stress models on human-derived neuronal networks and found each produced different patterns of neuronal activity and cell damage. The models tested were hypoxia alone, oxygen-glucose deprivation (OGD), and hypoxia combined with glutamate exposure. The findings suggest that choice of experimental model significantly affects results and that combining electrophysiological and structural analyses is important for accurately assessing metabolic stress in stroke research.