Picture for Yixiao Huang

Yixiao Huang

Fast Adversarial Training against Sparse Attacks Requires Loss Smoothing

Add code
Feb 28, 2025
Viaarxiv icon

On the Power of Convolution Augmented Transformer

Add code
Jul 08, 2024
Viaarxiv icon

Towards Efficient Training and Evaluation of Robust Models against $l_0$ Bounded Adversarial Perturbations

Add code
May 08, 2024
Viaarxiv icon

Mechanics of Next Token Prediction with Self-Attention

Add code
Mar 12, 2024
Figure 1 for Mechanics of Next Token Prediction with Self-Attention
Figure 2 for Mechanics of Next Token Prediction with Self-Attention
Figure 3 for Mechanics of Next Token Prediction with Self-Attention
Figure 4 for Mechanics of Next Token Prediction with Self-Attention
Viaarxiv icon

From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers

Add code
Feb 21, 2024
Figure 1 for From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers
Figure 2 for From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers
Figure 3 for From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers
Figure 4 for From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers
Viaarxiv icon