An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution

卷积神经网络的一个有趣的弱点与CoordConv解.

Inception Convolution with Efficient Dilation Search

高效扩张率搜索的Inception卷积.

Large Language Models Are Human-Level Prompt Engineers

大型语言模型是人类水平的提示工程师.

Scaling Instruction-Finetuned Language Models

扩展指令微调的语言模型.

What learning algorithm is in-context learning? Investigations with linear models

上下文学习算法是什么？使用线性模型的解释.

Inverse scaling can become U-shaped

逆向缩放可能会变成 U 形.