• Publications
  • Influence
An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
TLDR
We conduct a systematic evaluation of generic convolutional and recurrent architectures for sequence modeling. Expand
  • 832
  • 144
  • PDF
Multimodal Transformer for Unaligned Multimodal Language Sequences
TLDR
We introduce the Multimodal Transformer (MulT) to generically address the above issues in an end-to-end manner without explicitly aligning the data. Expand
  • 58
  • 14
  • PDF
Trellis Networks for Sequence Modeling
TLDR
We present trellis networks, a new architecture for sequence modeling that generalizes truncated recurrent networks with special structure, characterized by weight tying across depth and direct injection of the input into deep layers. Expand
  • 30
  • 4
  • PDF
Convolutional Sequence Modeling Revisited
TLDR
This paper looks at the problem of sequence modeling, predicting how a sequence will evolve over time. Expand
  • 20
  • 4
Deep Equilibrium Models
TLDR
We present a new approach to modeling sequential data: the deep equilibrium model (DEQ). Expand
  • 42
  • 3
  • PDF
Transformer Dissection: An Unified Understanding for Transformer's Attention via the Lens of Kernel
TLDR
In this paper, we present a new formulation of attention via the lens of the kernel. Expand
  • 19
  • 3
  • PDF
Multiscale Deep Equilibrium Models
TLDR
We propose a new class of implicit networks, the multiscale deep equilibrium model (MDEQ), suited to large-scale and highly hierarchical pattern recognition domains. Expand
  • 1
  • PDF
A community-powered search of machine learning strategy space to find NMR property prediction models
TLDR
We swarm search the space of ML strategies and develop algorithms for predicting atomic-pairwise nuclear magnetic resonance (NMR) properties in molecules. Expand
  • 2
  • PDF
Surfactant-assisted synthesis and luminescent properties study of LiGd(MoO4)2 phosphors
Abstract The LiGd(MoO4)2:Eu3+ phosphors were successfully synthesized using the hydrothermal method followed by calcination. The effects of different dosages of the cationic surfactant dodecylExpand