Transformer on a Diet

Reference: C Wang, Z Ye, A Zhang, Z Zhang, A Smola. "Transformer on a Diet". arXiv preprint arXiv (2020).

Installation

pip install --pre --upgrade mxnet
pip install gluonnlp

Results

The results and the command line to reproduce the results on PTB dataset are as follows.

[1] Full (Val PPL 109.19 Test PPL 103.72)

$ cd scripts/language_model/
$ python transformer_language_model.py --model full --data ptb --emsize 320 --nhid 2000 --nlayers 3 --lr 10 --epochs 500 --batch_size 20 --bptt 70 --dropout 0.4 --dropout_h 0.25 --dropout_i 0 --dropout_e 0 --weight_drop 0 --tied --alpha 0 --beta 0 --lr_update_interval 100 --lr_update_factor 1 --num_heads 16 --scaled --units 320 --use_residual --max_src_length 1000 --warmup_steps 0 --first_window_size 1 --kernel_size 3 --d_base 2

[2] Dilated (Val PPL 115.67 Test PPL 110.92)

$ cd scripts/language_model/
$ python transformer_language_model.py --model dilated --data ptb --emsize 320 --nhid 2000 --nlayers 3 --lr 10 --epochs 500 --batch_size 20 --bptt 70 --dropout 0.4 --dropout_h 0.25 --dropout_i 0 --dropout_e 0 --weight_drop 0 --tied --alpha 0 --beta 0 --lr_update_interval 100 --lr_update_factor 1 --num_heads 16 --scaled --units 320 --use_residual --max_src_length 1000 --warmup_steps 0 --first_window_size 1 --kernel_size 3 --d_base 2

[3] Dilated-Memory (Val PPL 115.35 Test PPL 110.98)

$ cd scripts/language_model/
$ python transformer_language_model.py --model dilated_mem --data ptb --emsize 320 --nhid 2000 --nlayers 3 --lr 10 --epochs 500 --batch_size 20 --bptt 70 --dropout 0.4 --dropout_h 0.25 --dropout_i 0 --dropout_e 0 --weight_drop 0 --tied --alpha 0 --beta 0 --lr_update_interval 100 --lr_update_factor 1 --num_heads 16 --scaled --units 320 --use_residual --max_src_length 1000 --warmup_steps 0 --first_window_size 1 --kernel_size 3 --d_base 2

[4] Cascade (Val PPL 109.16 Test PPL 105.27)

$ cd scripts/language_model/
$ python transformer_language_model.py --model cascade --data ptb --emsize 320 --nhid 2000 --nlayers 3 --lr 10 --epochs 500 --batch_size 20 --bptt 70 --dropout 0.4 --dropout_h 0.25 --dropout_i 0 --dropout_e 0 --weight_drop 0 --tied --alpha 0 --beta 0 --lr_update_interval 100 --lr_update_factor 1 --num_heads 16 --scaled --units 320 --use_residual --max_src_length 1000 --warmup_steps 0 --first_window_size 4 --window_size_multiplier 2 --kernel_size 3 --d_base 2

Note that the command to reproduce the results on wikitext-2 would be updated soon.

Reference Paper

The bibtext entry of the reference paper is:

@article{transformerdiet2020,
   title={Transformer on a Diet},
   author={Chenguang Wang and Zihao Ye and Aston Zhang and Zheng Zhang and Alexander J. Smola},
   journal={ArXiv},
   year={2020},
   volume={abs/2002.06170}
}

Code repo for "Transformer on a Diet" paper

Related tags

Overview

Transformer on a Diet

Installation

Results

Reference Paper

Owner

cgraywang

Unofficial PyTorch implementation of MobileViT.

Multi-Anchor Active Domain Adaptation for Semantic Segmentation (ICCV 2021 Oral)

Source code for our EMNLP'21 paper 《Raise a Child in Large Language Model: Towards Effective and Generalizable Fine-tuning》

[NeurIPS 2019] Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss

Adversarial Learning for Modeling Human Motion

VQGAN+CLIP Colab Notebook with user-friendly interface.

Official Code for "Non-deep Networks"

Geometric Deep Learning Extension Library for PyTorch

An efficient and effective learning to rank algorithm by mining information across ranking candidates. This repository contains the tensorflow implementation of SERank model. The code is developed based on TF-Ranking.

Benchmark for evaluating open-ended generation

NL-Augmenter 🦎 → 🐍 A Collaborative Repository of Natural Language Transformations

Pytorch implementation of the paper: "SAPNet: Segmentation-Aware Progressive Network for Perceptual Contrastive Image Deraining"

Direct LiDAR Odometry: Fast Localization with Dense Point Clouds

PyTorch EO aims to make Deep Learning for Earth Observation data easy and accessible to real-world cases and research alike.

✔️ Visual, reactive testing library for Julia. Time machine included.

Code for pre-training CharacterBERT models (as well as BERT models).

Official implementation of "StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation" (SIGGRAPH 2021)

The official repo for OC-SORT: Observation-Centric SORT on video Multi-Object Tracking. OC-SORT is simple, online and robust to occlusion/non-linear motion.

Pytorch implementation of forward and inverse Haar Wavelets 2D

Visual odometry package based on hardware-accelerated NVIDIA Elbrus library with world class quality and performance.