PyTea: PyTorch Tensor shape error analyzer

Overview

PyTea: PyTorch Tensor Shape Error Analyzer

paper project page

Requirements

  • node.js >= 12.x
  • python >= 3.8
    • z3-solver >= 4.8

How to install and use

# install node.js
sudo apt-get install nodejs

# install python z3-solver
pip install z3-solver

# download pytea
wget https://github.com/ropas/pytea/releases/download/v0.1.0/pytea.zip
unzip pytea.zip

# run pytea
python bin/pytea.py path/to/source.py

# run example file
python bin/pytea.py packages/pytea/pytest/basics/scratch.py

How to build

# install dependencies
npm run install:all
pip install z3-solver

# build
npm run build

Documentations

Brief explanation of the analysis result

PyTea is composed of two analyzers.

  • Online analysis: node.js (TypeScript / JavaScript)
    • Find numeric range-based shape mismatch and misuse of API argument. If PyTea has found any error while analyzing the code, it will stop at that position and inform the errors and violated constraints to the user.
  • Offline analysis: Z3 / Python
    • The generated constraints are passed to Z3Py. Z3 will solve the constraint sets of each path and print the first violated constraint (if it exists).

The result of the Online analyzer is divided into three classes:

  • potential success path: the analyzer does not found shape mismatch until now, but the final constraint set can be violated if Z3 analyzes it on closer inspection.
  • potential unreachable path: the analyzer found a shape mismatch or API misuses, but there remain path constraints. In short, path constraint is an unresolved branch condition; that means the stopped path might be unreachable if remaining path constraints have a contradiction. Those cases will be distinguished from Offline analysis.
  • immediate failed path: the analyzer has found an error, stops its analysis immediately.

CAVEAT: If the code contains PyTorch or other third-party APIs that we have not implemented, it will raise false alarms. Nevertheless, we also record each unimplemented API call. See LOGS section from the result and search which unimplemented API call is performed.

The final result of the Offline analysis is divided into several cases.

  • Valid path: SMT solver has not found any error. Every constraint will always be fulfilled.
  • Invalid path: SMT solver found a condition that can violate some constraints. Notice that this does not mean the code will always crash, but it found an extreme case that crashes some executions.
  • Undecidable path: SMT solver has met unsolvable constraints, then timeouts. Some non-linear formulae can be classified into this case.
  • Unreachable path: Hard and Path constraints contain contradicting constraints; this path will not be realized from the beginning.

Result examples

  • Error found by Online analysis

test1

  • Error found by Offline analysis

test2

License

MIT License

This project is based on Pyright, also MIT License

Comments
  • LSTM/GRU input_size tensor shape errors

    LSTM/GRU input_size tensor shape errors

    Hi, I have met a problem while detecting LSTM tensor shape errors. The testing file below is runnable and pytea returns correctly.

    import torch
    import torch.nn as nn
    import torch.nn.functional as F
    
    
    class Model(nn.Module):
    
        def __init__(self):
            super().__init__()
            self.cnn = nn.Conv2d(3, 1, 3, 1, 1)
            self.rnn = nn.LSTM(32, 64, 1, batch_first=True)
            self.pool = nn.MaxPool2d(2, 2)
            self.fc = nn.Linear(64, 16)
    
        def forward(self, x):
            x = self.pool(F.relu(self.cnn(x)))
            x = x.view(-1, 32, 32)
            x, _ = self.rnn(x)
            x = x[:, -1, :].squeeze(1)
            x = F.relu(self.fc(x))
            x = F.softmax(x, dim=-1)
            return x
    
    
    if __name__ == "__main__":
        net = Model()
        x = torch.randn(2, 3, 64, 64)
        y = net(x)
        target = torch.argmax(torch.randn(2, 16), dim=-1)
        loss = F.cross_entropy(y, target.long())
        loss.backward()
        print(y.size())
    

    However, If I change self.rnn = nn.LSTM(32, 64, 1, batch_first=True) into self.rnn = nn.LSTM(64, 64, 1, batch_first=True), torch will report a RuntimeError: Expected 64, got 32. pytea didn't return any CONSTRAINTS information, as it supposed to.

    Then I tried to more LSTM input_size shape errors, all failed. Same situation with GRU. I think it is a bug, because I can detect Conv2d, Linear error successfully.

    opened by MCplayerFromPRC 4
  • Inconsistent with document description

    Inconsistent with document description

    Hello, I have encountered the following problems:

    First question: The content of my source file is:

       import torch
       import torch.nn as nn
    
      class Net(nn.Module):
          def __init__(self):
              super(Net, self).__init__()
              self.layers = nn.Sequential(
                  nn.Linear(28 * 28, 120),
                  nn.ReLU(),
                  nn.Linear(80, 10))
      
          def a(self):
              pass
    
        if __name__ == "__main__":
            n = Net()
    

    But when I execute the command, I get the following results:

    image

    There should be a problem with defining shape in this model.

    Second question: I used it https://github.com/pytorch/examples/blob/master/mnist/main.py , but the command is stuck and no result is returned. As follows:

    image
    opened by dejavu6 2
  • Ternary expression (A if B else C) bug report

    Ternary expression (A if B else C) bug report

    아래와 같은 코드 실행에서 문제가 발생한다는 것을 깨달았습니다.

    x = 0 if 0 <= 1 else 1
    
    # runtime output
    REDUCED HEAP: (size: 250)
      x => 1
    

    파이썬의 삼항연산자가 x = (((0 <= 1) and 0) or 1)로 파싱됩니다. Logical statement가 True, true-value가 0일 때 발생하는 오류인 것으로 보입니다.

    당장 벤치마크 코드에서 나타나는 문제는 아닙니다. Pylib builtin 구현에서 발생한 문제이므로, 다른 방식으로 구현함으로써 일단은 피해갈 수 있을 것 같습니다.

    감사합니다.

    opened by lego0901 1
  • Develop sehoon

    Develop sehoon

    UT-1~6 코드 분석에 필요한 torch API 구현 완료.

    UT-2: epoch을 1로 수정하지 않으면 timeout이됨. UT-3: Python 빌트인 함수 iter, next의 구현은 우선 넘어갔음. UT-6: buggy 코드에서 target이 free variable인데, 이를 처리해 주지 않고 분석을 실행하면 아무것도 출력하지 않는 버그가 있음.

    위 특이사항을 적절히 처리해주고 분석을 실행하면 1~6 모두 buggy 코드는 invalid, fix 코드는 valid 결과를 냄.

    opened by Sehun0819 0
  • path constraint check

    path constraint check

    분석중 한 패스에서 (텐서 모양 오류 등의) 에러를 만나면 해당 패스는 처리됨. 문제는 분기 조건문에 의해 실제로는 진행되지 않는 패스여도 로 처리가 되는 것.

    따라서 분석중 에러를 만났을 때 그 패스가 path constraint를 갖고 있으면 로 처리하여 z3단에 넘기게 수정하였음.

    TODO: z3단에 넘기기 전에 path constraint를 계산하여 Valid면 , Unsat이면 로 처리하기(else )

    opened by Sehun0819 0
  • Bump node-notifier from 8.0.0 to 8.0.1 in /packages/pyright-internal

    Bump node-notifier from 8.0.0 to 8.0.1 in /packages/pyright-internal

    Bumps node-notifier from 8.0.0 to 8.0.1.

    Changelog

    Sourced from node-notifier's changelog.

    v8.0.1

    • fixes possible injection issue for notify-send
    Commits

    Dependabot compatibility score

    Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.


    Dependabot commands and options

    You can trigger Dependabot actions by commenting on this PR:

    • @dependabot rebase will rebase this PR
    • @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
    • @dependabot merge will merge this PR after your CI passes on it
    • @dependabot squash and merge will squash and merge this PR after your CI passes on it
    • @dependabot cancel merge will cancel a previously requested merge and block automerging
    • @dependabot reopen will reopen this PR if it is closed
    • @dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually
    • @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
    • @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
    • @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
    • @dependabot use these labels will set the current labels as the default for future PRs for this repo and language
    • @dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language
    • @dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language
    • @dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

    You can disable automated security fix PRs for this repo from the Security Alerts page.

    dependencies 
    opened by dependabot[bot] 0
Releases(v0.1.0)
Owner
ROPAS Lab.
ROPAS Lab. @ Seoul National University
ROPAS Lab.
PyTorch implementation of PSPNet

PSPNet with PyTorch Unofficial implementation of "Pyramid Scene Parsing Network" (https://arxiv.org/abs/1612.01105). This repository is just for caffe

Kazuto Nakashima 52 Nov 16, 2022
The codes and related files to reproduce the results for Image Similarity Challenge Track 1.

ISC-Track1-Submission The codes and related files to reproduce the results for Image Similarity Challenge Track 1. Required dependencies To begin with

Wenhao Wang 115 Jan 02, 2023
PyTorch implementations of algorithms for density estimation

pytorch-flows A PyTorch implementations of Masked Autoregressive Flow and some other invertible transformations from Glow: Generative Flow with Invert

Ilya Kostrikov 546 Dec 05, 2022
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks

StackGAN Pytorch implementation Inception score evaluation StackGAN-v2-pytorch Tensorflow implementation for reproducing main results in the paper Sta

Han Zhang 1.8k Dec 21, 2022
Codebase for Attentive Neural Hawkes Process (A-NHP) and Attentive Neural Datalog Through Time (A-NDTT)

Introduction Codebase for the paper Transformer Embeddings of Irregularly Spaced Events and Their Participants. This codebase contains two packages: a

Alan Yang 28 Dec 12, 2022
Source code for NAACL 2021 paper "TR-BERT: Dynamic Token Reduction for Accelerating BERT Inference"

TR-BERT Source code and dataset for "TR-BERT: Dynamic Token Reduction for Accelerating BERT Inference". The code is based on huggaface's transformers.

THUNLP 37 Oct 30, 2022
Self-supervised learning on Graph Representation Learning (node-level task)

graph_SSL Self-supervised learning on Graph Representation Learning (node-level task) How to run the code To run GRACE, sh run_GRACE.sh To run GCA, sh

Namkyeong Lee 3 Dec 31, 2021
The trained model and denoising example for paper : Cardiopulmonary Auscultation Enhancement with a Two-Stage Noise Cancellation Approach

The trained model and denoising example for paper : Cardiopulmonary Auscultation Enhancement with a Two-Stage Noise Cancellation Approach

ycj_project 1 Jan 18, 2022
机器学习、深度学习、自然语言处理等人工智能基础知识总结。

说明 机器学习、深度学习、自然语言处理基础知识总结。 目前主要参考李航老师的《统计学习方法》一书,也有一些内容例如XGBoost、聚类、深度学习相关内容、NLP相关内容等是书中未提及的。

Peter 445 Dec 12, 2022
Multi-Task Learning as a Bargaining Game

Nash-MTL Official implementation of "Multi-Task Learning as a Bargaining Game". Setup environment conda create -n nashmtl python=3.9.7 conda activate

Aviv Navon 87 Dec 26, 2022
Chinese Advertisement Board Identification(Pytorch)

Chinese-Advertisement-Board-Identification. We use YoloV5 to extract the ROI of the location of the chinese word. Next, we sort the bounding box and recognize every chinese words which we extracted.

Li-Wei Hsiao 12 Jul 21, 2022
Easy and Efficient Object Detector

EOD Easy and Efficient Object Detector EOD (Easy and Efficient Object Detection) is a general object detection model production framework. It aim on p

381 Jan 01, 2023
TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-Captured Scenarios

TPH-YOLOv5 This repo is the implementation of "TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-Captured

cv516Buaa 439 Dec 22, 2022
Reducing Information Bottleneck for Weakly Supervised Semantic Segmentation (NeurIPS 2021)

Reducing Information Bottleneck for Weakly Supervised Semantic Segmentation (NeurIPS 2021) The implementation of Reducing Infromation Bottleneck for W

Jungbeom Lee 81 Dec 16, 2022
Implementation detail for paper "Multi-level colonoscopy malignant tissue detection with adversarial CAC-UNet"

Multi-level-colonoscopy-malignant-tissue-detection-with-adversarial-CAC-UNet Implementation detail for our paper "Multi-level colonoscopy malignant ti

CVSM Group - email: <a href=[email protected]"> 84 Nov 22, 2022
Retinal Vessel Segmentation with Pixel-wise Adaptive Filters (ISBI 2022)

Retinal Vessel Segmentation with Pixel-wise Adaptive Filters (ISBI 2022) Introdu

anonymous 14 Oct 27, 2022
A PyTorch implementation of "SimGNN: A Neural Network Approach to Fast Graph Similarity Computation" (WSDM 2019).

SimGNN ⠀⠀⠀ A PyTorch implementation of SimGNN: A Neural Network Approach to Fast Graph Similarity Computation (WSDM 2019). Abstract Graph similarity s

Benedek Rozemberczki 534 Dec 25, 2022
code for paper "Not All Unlabeled Data are Equal: Learning to Weight Data in Semi-supervised Learning" by Zhongzheng Ren*, Raymond A. Yeh*, Alexander G. Schwing.

Not All Unlabeled Data are Equal: Learning to Weight Data in Semi-supervised Learning Overview This code is for paper: Not All Unlabeled Data are Equa

Jason Ren 22 Nov 23, 2022
😇A pyTorch implementation of the DeepMoji model: state-of-the-art deep learning model for analyzing sentiment, emotion, sarcasm etc

------ Update September 2018 ------ It's been a year since TorchMoji and DeepMoji were released. We're trying to understand how it's being used such t

Hugging Face 865 Dec 24, 2022
Pydantic models for pywttr and aiopywttr.

Pydantic models for pywttr and aiopywttr.

Almaz 2 Dec 08, 2022