ConvMAE: Masked Convolution Meets Masked Autoencoders

ConvMAE：结合掩码卷积与掩码自编码器.

SimMIM: A Simple Framework for Masked Image Modeling

SimMIM：一种掩码图像建模的简单框架.

iBOT: Image BERT Pre-Training with Online Tokenizer

iBOT：使用在线标志进行图像BERT预训练.

BEiT: BERT Pre-Training of Image Transformers

BEiT：图像Transformer中的BERT预训练.

Discovering faster matrix multiplication algorithms with reinforcement learning

AlphaTensor：通过强化学习发现更快的矩阵乘法算法.

Analyzing and Improving Representations with the Soft Nearest Neighbor Loss

通过软最近邻损失分析和改进表示学习.