请问原始GoogleNet（Inception）论文中采用的batch size是多少？

After reviewing the methodology section of the original GoogleNet (Inception) paper, I can confirm that the batch size parameter is not explicitly stated within that document.

If you're aiming to replicate the model's training setup, you may want to look towards widely adopted open-source implementations of GoogleNet (which often use batch sizes like 32 or 64 as common defaults) or follow-up publications that reference the original work's training details. However, the core paper itself doesn't include this specific parameter.

内容的提问来源于stack exchange，提问作者pecorino