• Stars
    star
    473
  • Rank 92,832 (Top 2 %)
  • Language
    Python
  • License
    MIT License
  • Created about 6 years ago
  • Updated about 5 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

[ICLR'19] Trellis Networks for Sequence Modeling

TrellisNet for Sequence Modeling

PWC PWC

This repository contains the experiments done in paper Trellis Networks for Sequence Modeling by Shaojie Bai, J. Zico Kolter and Vladlen Koltun.

On the one hand, a trellis network is a temporal convolutional network with special structure, characterized by weight tying across depth and direct injection of the input into deep layers. On the other hand, we show that truncated recurrent networks are equivalent to trellis networks with special sparsity structure in their weight matrices. Thus trellis networks with general weight matrices generalize truncated recurrent networks. This allows trellis networks to serve as bridge between recurrent and convolutional architectures, benefitting from algorithmic and architectural techniques developed in either context. We leverage these relationships to design high-performing trellis networks that absorb ideas from both architectural families. Experiments demonstrate that trellis networks outperform the current state of the art on a variety of challenging benchmarks, including word-level language modeling on Penn Treebank and WikiText-103 (UPDATE: recently surpassed by Transformer-XL), character-level language modeling on Penn Treebank, and stress tests designed to evaluate long-term memory retention.

Our experiments were done in PyTorch. If you find our work, or this repository helpful, please consider citing our work:

@inproceedings{bai2018trellis,
  author    = {Shaojie Bai and J. Zico Kolter and Vladlen Koltun},
  title     = {Trellis Networks for Sequence Modeling},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2019},
}

Datasets

The code should be directly runnable with PyTorch 1.0.0 or above. This repository contains the training script for the following tasks:

  • Sequential MNIST handwritten digit classification
  • Permuted Sequential MNIST that randomly permutes the pixel order in sequential MNIST
  • Sequential CIFAR-10 classification (more challenging, due to more intra-class variations, channel complexities and larger images)
  • Penn Treebank (PTB) word-level language modeling (with and without the mixture of softmax); vocabulary size 10K
  • Wikitext-103 (WT103) large-scale word-level language modeling; vocabulary size 268K
  • Penn Treebank medium-scale character-level language modeling

Note that these tasks are on very different scales, with unique properties that challenge sequence models in different ways. For example, word-level PTB is a small dataset that a typical model easily overfits, so judicious regularization is essential. WT103 is a hundred times larger, with less danger of overfitting, but with a vocabulary size of 268K that makes training more challenging (due to large embedding size).

Pre-trained Model(s)

We provide some reasonably good pre-trained weights here so that the users don't need to train from scratch. We'll update the table from time to time. (Note: if you train from scratch using different seeds, it's likely you will get better results :-))

Description Task Dataset Model
TrellisNet-LM Word-Level Language Modeling Penn Treebank (PTB) download (.pkl)
TrellisNet-LM Character-Level Language Modeling Penn Treebank (PTB) download (.pkl)

To use the pre-trained weights, use the flag --load_weight [.pkl PATH] when starting the training script (e.g., you can just use the default arg parameters). You can use the flag --eval turn on the evaluation mode only.

Usage

All tasks share the same underlying TrellisNet model, which is in file trellisnet.py (and the eventual models, including components like embedding layer, are in model.py). As discussed in the paper, TrellisNet is able to benefit significantly from techniques developed originally for RNNs as well as temporal convolutional networks (TCNs). Some of these techniques are also included in this repository. Each task is organized in the following structure:

[TASK_NAME] /
    data/
    logs/
    [TASK_NAME].py
    model.py
    utils.py
    data.py

where [TASK_NAME].py is the training script for the task (with argument flags; use -h to see the details).

More Repositories

1

TCN

Sequence modeling benchmarks and temporal convolutional networks
Python
4,122
star
2

convmixer

Implementation of ConvMixer for "Patches Are All You Need? 🀷"
Python
1,059
star
3

mpc.pytorch

A fast and differentiable model predictive control (MPC) solver for PyTorch.
Python
865
star
4

deq

[NeurIPS'19] Deep Equilibrium Models
Python
719
star
5

qpth

A fast and differentiable QP solver for PyTorch.
Python
673
star
6

wanda

A simple and effective LLM pruning approach.
Python
602
star
7

optnet

OptNet: Differentiable Optimization as a Layer in Neural Networks
Python
507
star
8

fast_adversarial

[ICLR 2020] A repository for extremely fast adversarial training using FGSM
Python
422
star
9

SATNet

Bridging deep learning and logical reasoning using a differentiable satisfiability solver.
Python
404
star
10

convex_adversarial

A method for training neural networks that are provably robust to adversarial attacks.
Python
378
star
11

smoothing

Provable adversarial robustness at ImageNet scale
Python
357
star
12

pytorch_fft

PyTorch wrapper for FFTs
Python
313
star
13

lcp-physics

A differentiable LCP physics engine in PyTorch.
Python
292
star
14

icnn

Input Convex Neural Networks
Python
274
star
15

differentiable-mpc

Python
239
star
16

mdeq

[NeurIPS'20] Multiscale Deep Equilibrium Models
Python
232
star
17

e2e-model-learning

Task-based end-to-end model learning in stochastic optimization
Python
195
star
18

ect

Consistency Models Made Easy
Python
188
star
19

deq-flow

[CVPR 2022] Deep Equilibrium Optical Flow Estimation
Python
177
star
20

robust_overfitting

Python
153
star
21

DC3

DC3: A Learning Method for Optimization with Hard Constraints
Python
133
star
22

cfd-gcn

Python
113
star
23

massive-activations

Code accompanying the paper "Massive Activations in Large Language Models"
Python
95
star
24

tofu

Landing Page for TOFU
Python
86
star
25

FLYP

Code for Finetune like you pretrain: Improved finetuning of zero-shot vision models
Python
85
star
26

projected_sinkhorn

Python
85
star
27

torchdeq

Modern Fixed Point Systems using Pytorch
Python
74
star
28

perturbation_learning

Learning perturbation sets for robust machine learning
Python
64
star
29

scaling_laws_data_filtering

Python
59
star
30

lml

The Limited Multi-Label Projection Layer
Python
58
star
31

deq-ddim

Python
58
star
32

chatllm-vscode

TypeScript
58
star
33

edge-of-stability

Python
55
star
34

robust-nn-control

Enforcing robust control guarantees within neural network policies
Python
52
star
35

monotone_op_net

Monotone operator equilibrium networks
Jupyter Notebook
51
star
36

orthogonal-convolutions

Implementations of orthogonal and semi-orthogonal convolutions in the Fourier domain with applications to adversarial robustness
Jupyter Notebook
41
star
37

convmixer-cifar10

Simple CIFAR-10 classification with ConvMixer
Python
40
star
38

newton_admm

A Newton ADMM based solver for Cone programming.
Python
38
star
39

tta_conjugate

Test-Time Adaptation via Conjugate Pseudo-Labels
Python
36
star
40

T-MARS

Code for T-MARS data filtering
Python
34
star
41

stable_dynamics

Companion code to "Learning Stable Deep Dynamics Models" (Manek and Kolter, 2019)
Jupyter Notebook
31
star
42

ImpSq

Implicit^2: Implicit model for implicit neural representations
Python
27
star
43

robust_union

[ICML'20] Multi Steepest Descent (MSD) for robustness against the union of multiple perturbation models.
Python
25
star
44

breaking-poisoned-classifier

Code for paper "Poisoned classifiers are not only backdoored, they are fundamentally broken"
Jupyter Notebook
24
star
45

diffusion-model-hallucination

Python
24
star
46

acr-memorization

Python
24
star
47

gradient_regularized_gan

Code for "Gradient descent GAN optimization is locally stable"
Python
21
star
48

get

Generative Equilibrium Transformer
Python
17
star
49

smoothinv

Single Image Backdoor Inversion via Robust Smoothed Classifiers
Python
16
star
50

intermediate_robustness

Python
16
star
51

mixing

The Mixing method: coordinate descent for low-rank semidefinite programming
C
15
star
52

dreaml

dreaml: dynamic reactive machine learning
JavaScript
12
star
53

ase

Analogous Safe-state Exploration (ASE) is an algorithm for provably safe and optimal exploration in MDPs with unknown, stochastic dynamics.
Python
11
star
54

sdp_clustering

Jupyter Notebook
11
star
55

JIIO-DEQ

Efficient joint input optimization and inference with DEQ
Python
10
star
56

uniform-convergence-NeurIPS19

The code for the NeurIPS19 paper and blog on "Uniform convergence may be unable to explain generalization in deep learning".
Jupyter Notebook
10
star
57

sdp_mrf

Jupyter Notebook
3
star
58

mixsat

Low-rank semidefinite programming for the MAX2SAT problem
C
3
star
59

MonotoneDBM

Python
2
star
60

lipschitz_mondeq

Jupyter Notebook
1
star
61

mugrade

Python
1
star