Search papers, labs, and topics across Lattice.
National Uni- versity of Singapore, Faculty of Computer Science and Engineering, Ho Chi Minh City University of Technology (HC- MUT), VNU-HCM
3
0
3
Lower-order discontinuities in Sparse Mixture-of-Experts architectures dominate input space, significantly affecting model outputs and performance.
Conservation laws reveal how modern neural architectures maintain implicit biases during training, shedding light on their remarkable performance.
Rotary positional encodings not only enhance expressivity in Transformers but also fundamentally alter the symmetry landscape of attention mechanisms.
