tech-srl / layer_norm_expressivity_roleLinks
Code for the paper "On the Expressivity Role of LayerNorm in Transformers' Attention" (Findings of ACL'2023)
☆56Updated 11 months ago
Alternatives and similar repositories for layer_norm_expressivity_role
Users that are interested in layer_norm_expressivity_role are comparing it to the libraries listed below
Sorting:
- Using FlexAttention to compute attention with different masking patterns☆44Updated last year
- ☆51Updated last year
- Triton Implementation of HyperAttention Algorithm☆48Updated last year
- Official repository of paper "RNNs Are Not Transformers (Yet): The Key Bottleneck on In-context Retrieval"☆27Updated last year
- Code for NeurIPS 2024 Spotlight: "Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations"☆83Updated 10 months ago
- Source-to-Source Debuggable Derivatives in Pure Python☆15Updated last year
- ☆56Updated last year
- ☆38Updated last year
- Explorations into the recently proposed Taylor Series Linear Attention☆100Updated last year
- Official code for the paper "Attention as a Hypernetwork"☆42Updated last year
- ☆85Updated last year
- ☆100Updated last year
- Sequence Modeling with Multiresolution Convolutional Memory (ICML 2023)☆127Updated last year
- Official PyTorch implementation and models for paper "Diffusion Beats Autoregressive in Data-Constrained Settings". We find diffusion mod…☆92Updated 3 weeks ago
- Implementation of Token Shift GPT - An autoregressive model that solely relies on shifting the sequence space for mixing☆50Updated 3 years ago
- Code for the paper "Function-Space Learning Rates"☆23Updated 3 months ago
- Here we will test various linear attention designs.☆62Updated last year
- ☆29Updated 2 years ago
- Xmixers: A collection of SOTA efficient token/channel mixers☆14Updated 2 weeks ago
- Implementation of Infini-Transformer in Pytorch☆111Updated 8 months ago
- The official repository for our paper "The Dual Form of Neural Networks Revisited: Connecting Test Time Predictions to Training Patterns …☆16Updated 3 months ago
- Why Do We Need Weight Decay in Modern Deep Learning? [NeurIPS 2024]☆67Updated 11 months ago
- Deep Networks Grok All the Time and Here is Why☆37Updated last year
- GoldFinch and other hybrid transformer components☆45Updated last year
- ☆45Updated last year
- One Initialization to Rule them All: Fine-tuning via Explained Variance Adaptation☆42Updated 11 months ago
- Experimental scripts for researching data adaptive learning rate scheduling.☆22Updated last year
- [ICML 24 NGSM workshop] Associative Recurrent Memory Transformer implementation and scripts for training and evaluation☆52Updated 2 weeks ago
- Code for the paper "Query-Key Normalization for Transformers"☆48Updated 4 years ago
- Linear Attention Sequence Parallelism (LASP)☆86Updated last year