Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Towards Better & Faster Autoregressive Image Generation: From the Perspective of Entropy

Authors: Xiaoxiao Ma, Feng Zhao, Pengyang Ling, Haibo Qiu, Zhixiang Wei, Hu Yu, Jie Huang, Zhixiong Zeng, Lin Ma

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments across multiple benchmarks using diverse AR image generation models demonstrate the effectiveness and generalizability of our approach in enhancing both generation quality and sampling speed.
Researcher Affiliation Collaboration 1University of Science and Technology of China 2Meituan
Pseudocode No The paper describes the methods using mathematical equations and textual explanations, but no explicitly labeled pseudocode or algorithm blocks are present.
Open Source Code Yes Code is available at https://github.com/krennic999/ARsample.
Open Datasets Yes FID and CLIP-Score are tested on the MS-COCO 2017 [64] validation set... To assess the potential misjudgment of model performance caused by limited image numbers, we further evaluated the metrics on a larger dataset, COCO2014, as shown in Table 6.
Dataset Splits Yes FID and CLIP-Score are tested on the MS-COCO 2017 [64] validation set to evaluate the image quality and prompt-following capability.
Hardware Specification Yes All experiments are conducted on A100 GPUs.
Software Dependencies No The paper does not explicitly list specific software dependencies with version numbers (e.g., Python, PyTorch, CUDA versions) that would be needed to replicate the experiments.
Experiment Setup Yes All models are evaluated under their original inference settings (e.g., CFG=4 and top K=2000 for Lumina-m GPT, CFG=7.5 for Llama Gen). In Sec. 3.2, we propose to dynamically control the sampling temperature based on entropy. [...] we list the detailed parameters for each model in Table 5.