苹果的研究人员近日推出了一种名为AIM的自回归视觉模型。他们在最新论文《Scalable Pre-training of Large Autoregressive Image Models》中提出，通过自回归目标训练ViT模型，可以在学习表征方面获得与大型语言模型（LLMs）相同的扩展能力。研究结果显示，模型容量可以轻松扩展到数十亿个参数，并且AIM能有效利用大量未经整理的图像数据。
据“机器之心”报道，这一突破性成果进一步推动了计算机视觉领域的发展。苹果研究团队通过大量实验验证了AIM模型的有效性，并在多个图像识别任务上取得了优异的性能表现。据悉，AIM模型在图像分类、目标检测和图像生成等任务上均取得了较好的成绩。
苹果公司一直以来都在人工智能领域进行深入研究，并取得了诸多重要成果。此次发布的AIM模型不仅在计算机视觉领域引起了广泛关注，也为图像处理和人工智能技术的发展提供了新的思路。未来，有望看到更多基于AIM模型创新应用的出现。
Title: Apple Releases Autoregressive Visual Model AIM
Keywords: Apple, Autoregressive Visual Model, AIM
News content:
Apple researchers have recently introduced an autoregressive visual model called AIM. In their latest paper “Scalable Pre-training of Large Autoregressive Image Models,” they propose that training ViT models with autoregressive objectives can achieve the same scalability in learning representations as large language models (LLMs). The research results show that the model capacity can easily expand to billions of parameters, and AIM can effectively utilize a large amount of unstructured image data.
According to a report by “Machine Heart,” this breakthrough has further promoted the development of the field of computer vision. The Apple research team has validated the effectiveness of the AIM model through a large number of experiments and achieved excellent performance in multiple image recognition tasks. It is reported that the AIM model has achieved good results in tasks such as image classification, object detection, and image generation.
Apple has always been deeply involved in artificial intelligence research and has achieved many important results. The release of the AIM model has not only attracted widespread attention in the field of computer vision but also provided new ideas for the development of image processing and artificial intelligence technology. In the future, it is expected to see more innovative applications based on the AIM model.
【来源】https://www.jiqizhixin.com/articles/2024-01-18-7
Views: 1