SMT-Based Translation Validation for Machine Learning Compiler

被引:7
|
作者
Bang, Seongwon [1 ]
Nam, Seunghyeon [1 ]
Chun, Inwhan [1 ]
Jhoo, Ho Young [1 ]
Lee, Juneyoung [2 ]
机构
[1] Seoul Natl Univ, Seoul, South Korea
[2] CryptoLab, Seoul, South Korea
基金
新加坡国家研究基金会;
关键词
D O I
10.1007/978-3-031-13188-2_19
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
Machine learning compilers are large software containing complex transformations for deep learning models, and any buggy transformation may cause a crash or silently bring a regression to the prediction accuracy and performance. This paper proposes an SMT-based translation validation framework for Multi-Level IR (MLIR), a compiler framework used by many deep learning compilers. It proposes an SMT encoding tailored for translation validation that is an over-approximation of the FP arithmetic and reduction operations. It performs abstraction refinement if validation fails. We also propose a new approach for encoding arithmetic properties of reductions in SMT. We found mismatches between the specification and implementation of MLIR, and validated high-level transformations for SqueezeNet, MobileNet, and text_classification with proper splitting.
引用
收藏
页码:386 / 407
页数:22
相关论文
共 50 条
  • [1] SMT-Based Validation of Timed Failure Propagation Graphs
    Bozzano, Marco
    Cimatti, Alessandro
    Gario, Marco
    Micheli, Andrea
    PROCEEDINGS OF THE TWENTY-NINTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2015, : 3724 - 3730
  • [2] Dartagnan: SMT-based Violation Witness Validation (Competition Contribution)
    Ponce-de-Leon, Hernan
    Haas, Thomas
    Meyer, Roland
    TOOLS AND ALGORITHMS FOR THE CONSTRUCTION AND ANALYSIS OF SYSTEMS, TACAS 2022, PT II, 2022, 13244 : 418 - 423
  • [3] Ground Setting Properties for an Efficient Translation of OCL in SMT-based Model Finding
    Przigoda, Nils
    Wille, Robert
    Drechsler, Rolf
    19TH ACM/IEEE INTERNATIONAL CONFERENCE ON MODEL DRIVEN ENGINEERING LANGUAGES AND SYSTEMS (MODELS'16), 2016, : 261 - 271
  • [4] SMT-Based Verification of NGAC Policies
    Duhrovenski, Vladislav
    Chen, Erzhuo
    Xu, Dianxiang
    2023 IEEE 47TH ANNUAL COMPUTERS, SOFTWARE, AND APPLICATIONS CONFERENCE, COMPSAC, 2023, : 860 - 869
  • [5] An SMT-Based Approach to Coverability Analysis
    Esparza, Javier
    Ledesma-Garza, Ruslan
    Majumdar, Rupak
    Meyer, Philipp
    Niksic, Filip
    COMPUTER AIDED VERIFICATION, CAV 2014, 2014, 8559 : 603 - 619
  • [6] SMT-based generation of symbolic automata
    Xudong Qin
    Simon Bliudze
    Eric Madelaine
    Zechen Hou
    Yuxin Deng
    Min Zhang
    Acta Informatica, 2020, 57 : 627 - 656
  • [7] SMT-Based Software Model Checking
    Cimatti, Alessandro
    MODEL CHECKING SOFTWARE, 2010, 6349 : 1 - 3
  • [8] Integrating Machine Learning into an SMT-Based Planning Approach for Production Planning in Cyber-Physical Production Systems
    Heesch, Rene
    Ehrhardt, Jonas
    Niggemann, Oliver
    ARTIFICIAL INTELLIGENCE-ECAI 2023 INTERNATIONAL WORKSHOPS, PT 2, XAI3, TACTIFUL, XI-ML, SEDAMI, RAAIT, AI4S, HYDRA, AI4AI, 2023, 2024, 1948 : 318 - 331
  • [9] SMT-Based Variability Analyses in FeatureIDE
    Sprey, Joshua
    Sundermann, Chico
    Krieter, Sebastian
    Nieke, Michael
    Mauro, Jacopo
    Thiim, Thomas
    Schaefer, Ina
    PROCEEDINGS OF THE 14TH INTERNATIONAL WORKING CONFERENCE ON VARIABILITY MODELLING OF SOFTWARE-INTENSIVE SYSTEMS (VAMOS '20), 2020,
  • [10] SMT-Based Verification of Parameterized Systems
    Gurfinkel, Arie
    Shoham, Sharon
    Meshman, Yuri
    FSE'16: PROCEEDINGS OF THE 2016 24TH ACM SIGSOFT INTERNATIONAL SYMPOSIUM ON FOUNDATIONS OF SOFTWARE ENGINEERING, 2016, : 338 - 348