PyTorch implementation of Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation.

Overview

ALiBi

PyTorch implementation of Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation.

alt-text

Quickstart

Clone this repository.

git clone https://github.com/jaketae/alibi.git

Navigate to the cloned directory. You can use the bare-bone ALiBi decoder via

>>> import torch; from alibi import ALiBiConfig, ALiBiTransformer
>>> config  = ALiBiConfig()
>>> model = ALiBiTransformer(config)
>>> x = torch.randn(8, 100, 256)
>>> model(x).shape
torch.Size([8, 100, 256])

By default, the model comes with the following parameters:

ALiBiConfig(
    num_layers=6, 
    d_model=256, 
    num_heads=8, 
    max_len=256, 
    dropout=0.1, 
    causal=True, 
    expansion_factor=1
)

To use an encoder instead of a decoder, simply toggle causal=False.

Abstract

Since the introduction of the transformer model by Vaswani et al. (2017), a fundamental question remains open: how to achieve extrapolation at inference time to longer sequences than seen during training? We first show that extrapolation can be improved by changing the position representation method, though we find that existing proposals do not allow efficient extrapolation. We introduce a simple and efficient method, Attention with Linear Biases (ALiBi), that allows for extrapolation. ALiBi does not add positional embeddings to the word embeddings; instead, it biases the query-key attention scores with a term that is proportional to their distance. We show that this method allows training a 1.3 billion parameter model on input sequences of length 1024 that extrapolates to input sequences of length 2048, achieving the same perplexity as a sinusoidal position embedding model trained on inputs of length 2048, 11% faster and using 11% less memory. ALiBi's inductive bias towards recency allows it to outperform multiple strong position methods on the WikiText-103 benchmark. Finally, we provide analysis of ALiBi to understand why it leads to better performance.

Citation

@misc{press2021train,
	title        = {Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation},
	author       = {Ofir Press and Noah A. Smith and Mike Lewis},
	year         = 2021,
	eprint       = {2108.12409},
	archiveprefix = {arXiv},
	primaryclass = {cs.CL}
}
Owner
Jake Tae
CS + Math @ Yale, SWE intern @huggingface
Jake Tae
clustering moroccan stocks time series data using k-means with dtw (dynamic time warping)

Moroccan Stocks Clustering Context Hey! we don't always have to forecast time series am I right ? We use k-means to cluster about 70 moroccan stock pr

Ayman Lafaz 7 Oct 18, 2022
NU-Wave: A Diffusion Probabilistic Model for Neural Audio Upsampling @ INTERSPEECH 2021 Accepted

NU-Wave — Official PyTorch Implementation NU-Wave: A Diffusion Probabilistic Model for Neural Audio Upsampling Junhyeok Lee, Seungu Han @ MINDsLab Inc

MINDs Lab 242 Dec 23, 2022
Simple Tensorflow implementation of Toward Spatially Unbiased Generative Models (ICCV 2021)

Spatial unbiased GANs — Simple TensorFlow Implementation [Paper] : Toward Spatially Unbiased Generative Models (ICCV 2021) Abstract Recent image gener

Junho Kim 16 Apr 15, 2022
Open-source codebase for EfficientZero, from "Mastering Atari Games with Limited Data" at NeurIPS 2021.

EfficientZero (NeurIPS 2021) Open-source codebase for EfficientZero, from "Mastering Atari Games with Limited Data" at NeurIPS 2021. Thank you for you

Weirui Ye 671 Jan 03, 2023
Reproduction of Vision Transformer in Tensorflow2. Train from scratch and Finetune.

Vision Transformer(ViT) in Tensorflow2 Tensorflow2 implementation of the Vision Transformer(ViT). This repository is for An image is worth 16x16 words

sungjun lee 42 Dec 27, 2022
The implementation of 'Image synthesis via semantic composition'.

Image synthesis via semantic synthesis [Project Page] by Yi Wang, Lu Qi, Ying-Cong Chen, Xiangyu Zhang, Jiaya Jia. Introduction This repository gives

DV Lab 71 Jan 06, 2023
Implementation of a Transformer that Ponders, using the scheme from the PonderNet paper

Ponder(ing) Transformer Implementation of a Transformer that learns to adapt the number of computational steps it takes depending on the difficulty of

Phil Wang 65 Oct 04, 2022
Pytorch implementation for Patient Knowledge Distillation for BERT Model Compression

Patient Knowledge Distillation for BERT Model Compression Knowledge distillation for BERT model Installation Run command below to install the environm

Siqi 180 Dec 19, 2022
ICCV2021 Paper: AutoShape: Real-Time Shape-Aware Monocular 3D Object Detection

ICCV2021 Paper: AutoShape: Real-Time Shape-Aware Monocular 3D Object Detection

Zongdai 107 Dec 20, 2022
Vision-and-Language Navigation in Continuous Environments using Habitat

Vision-and-Language Navigation in Continuous Environments (VLN-CE) Project Website — VLN-CE Challenge — RxR-Habitat Challenge Official implementations

Jacob Krantz 132 Jan 02, 2023
A micro-game "flappy bird".

1-o-flappy A micro-game "flappy bird". Gameplays The game will be installed at /usr/bin . The name of it is "1-o-flappy". You can type "1-o-flappy" to

1 Nov 06, 2021
Official repo for BMVC2021 paper ASFormer: Transformer for Action Segmentation

ASFormer: Transformer for Action Segmentation This repo provides training & inference code for BMVC 2021 paper: ASFormer: Transformer for Action Segme

42 Dec 23, 2022
This PyTorch package implements MoEBERT: from BERT to Mixture-of-Experts via Importance-Guided Adaptation (NAACL 2022).

MoEBERT This PyTorch package implements MoEBERT: from BERT to Mixture-of-Experts via Importance-Guided Adaptation (NAACL 2022). Installation Create an

Simiao Zuo 34 Dec 24, 2022
(SIGIR2020) “Asymmetric Tri-training for Debiasing Missing-Not-At-Random Explicit Feedback’’

Asymmetric Tri-training for Debiasing Missing-Not-At-Random Explicit Feedback About This repository accompanies the real-world experiments conducted i

yuta-saito 19 Dec 01, 2022
Allows including an action inside another action (by preprocessing the Yaml file). This is how composite actions should have worked.

actions-includes Allows including an action inside another action (by preprocessing the Yaml file). Instead of using uses or run in your action step,

Tim Ansell 70 Nov 04, 2022
Deep Inertial Prediction (DIPr)

Deep Inertial Prediction For more information and context related to this repo, please refer to our website. Getting Started (non Docker) Note: you wi

Arcturus Industries 12 Nov 11, 2022
Self-Supervised Speech Pre-training and Representation Learning Toolkit.

What's New Sep 2021: We host a challenge in AAAI workshop: The 2nd Self-supervised Learning for Audio and Speech Processing! See SUPERB official site

s3prl 1.6k Jan 08, 2023
A resource for learning about deep learning techniques from regression to LSTM and Reinforcement Learning using financial data and the fitness functions of algorithmic trading

A tour through tensorflow with financial data I present several models ranging in complexity from simple regression to LSTM and policy networks. The s

195 Dec 07, 2022
Junction Tree Variational Autoencoder for Molecular Graph Generation (ICML 2018)

Junction Tree Variational Autoencoder for Molecular Graph Generation Official implementation of our Junction Tree Variational Autoencoder https://arxi

Wengong Jin 418 Jan 07, 2023
Analysis of Smiles through reservoir sampling & RDkit

Analysis of Smiles through reservoir sampling and machine learning (under development). This is a simple project that includes two Jupyter files for t

Aurimas A. Nausėdas 6 Aug 30, 2022