twistedcubic / attention-rank-collapse
[ICML 2021 Oral] We show pure attention suffers rank collapse, and how different mechanisms combat it.
☆164Updated 3 years ago
Alternatives and similar repositories for attention-rank-collapse:
Users that are interested in attention-rank-collapse are comparing it to the libraries listed below
- Official PyTorch Implementation of Long-Short Transformer (NeurIPS 2021).☆226Updated 2 years ago
- Official code repository of the paper Linear Transformers Are Secretly Fast Weight Programmers.☆102Updated 3 years ago
- Implementation of Long-Short Transformer, combining local and global inductive biases for attention over long sequences, in Pytorch☆118Updated 3 years ago
- Code for Multi-Head Attention: Collaborate Instead of Concatenate☆152Updated last year
- [ICML 2020] code for the flooding regularizer proposed in "Do We Need Zero Training Loss After Achieving Zero Training Error?"☆92Updated 2 years ago
- Awesome Contrastive Learning for CV & NLP☆162Updated 3 years ago
- [ICML 2020] code for "PowerNorm: Rethinking Batch Normalization in Transformers" https://arxiv.org/abs/2003.07845☆119Updated 3 years ago
- Transformer with Untied Positional Encoding (TUPE). Code of paper "Rethinking Positional Encoding in Language Pre-training". Improve exis…☆250Updated 3 years ago
- PyTorch implementation of a Variational Autoencoder with Gumbel-Softmax Distribution☆206Updated 6 years ago
- NeurIPS 2020, Debiased Contrastive Learning☆282Updated last year
- Implementation of Memformer, a Memory-augmented Transformer, in Pytorch☆110Updated 4 years ago
- Sinkhorn Transformer - Practical implementation of Sparse Sinkhorn Attention☆257Updated 3 years ago
- Apollo: An Adaptive Parameter-wise Diagonal Quasi-Newton Method for Nonconvex Stochastic Optimization☆182Updated 3 years ago
- [ICLR 2022] Official implementation of cosformer-attention in cosFormer: Rethinking Softmax in Attention☆185Updated 2 years ago
- ☆81Updated 6 months ago
- Fully featured implementation of Routing Transformer☆288Updated 3 years ago
- Open source code for paper "Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere" ICML 2…☆437Updated 2 years ago
- Understanding Training Dynamics of Deep ReLU Networks☆284Updated 2 weeks ago
- Implementation of Sparsemax activation in Pytorch☆158Updated 4 years ago
- ☆83Updated 3 years ago
- ☆172Updated 6 months ago
- Understanding the Difficulty of Training Transformers☆328Updated 2 years ago
- MoCo with Alignment and Uniformity Loss.☆61Updated 3 years ago
- Code for the ICML'20 paper "Improving Transformer Optimization Through Better Initialization"☆89Updated 4 years ago
- A Pytorch implementation of the optimal transport kernel embedding☆114Updated 4 years ago
- Code for Which Tasks Should Be Learned Together in Multi-task Learning?☆94Updated last year
- Template and style files for ICLR☆183Updated 6 months ago
- A supplementary code for Editable Neural Networks, an ICLR 2020 submission.☆45Updated 5 years ago
- Codes for CyGen, the novel generative modeling framework proposed in "On the Generative Utility of Cyclic Conditionals" (NeurIPS-21)☆45Updated 2 years ago
- Learning to Encode Position for Transformer with Continuous Dynamical Model☆59Updated 4 years ago