InterpretDL: Explaining Deep Models in PaddlePaddle

Authors: Xuhong Li, Haoyi Xiong, Xingjian Li, Xuanyu Wu, Zeyu Chen, Dejing Dou

JMLR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Interpret DL also provides a number of tutorial examples and showcases to demonstrate the capability of Interpret DL working on a wide range of deep learning models, e.g., Convolutional Neural Networks (CNNs), Multi-Layer Preceptors (MLPs), Transformers, etc., for various tasks in both Computer Vision (CV) and Natural Language Processing (NLP).
Researcher Affiliation Industry Xuhong Li EMAIL Haoyi Xiong EMAIL Xingjian Li EMAIL Xuanyu Wu v EMAIL Zeyu Chen EMAIL Dejing Dou EMAIL Baidu, Inc., Beijing, China
Pseudocode No The paper provides a code listing (Listing 1) as an example of how to use the Interpret DL library, but it does not contain structured pseudocode or algorithm blocks describing the methods within the paper.
Open Source Code Yes The project is available at https://github.com/Paddle Paddle/Interpret DL. Keywords: Explanation, Interpretation Algorithms, Trustworthiness, Deep Models
Open Datasets No The paper mentions using a "pretrained ResNet-50 model" and a "sentiment classifier based on texts" for its examples. While ResNet-50 is typically trained on ImageNet, and sentiment classifiers use common NLP datasets, the paper does not explicitly provide concrete access information (link, DOI, specific citation with authors/year) for the datasets used in its examples, nor does it explicitly state they are publicly available in the context of the experiments conducted.
Dataset Splits No The paper demonstrates the Interpret DL toolkit with examples of interpreting a pretrained ResNet-50 model and a sentiment classifier. It does not provide any specific details about the training, validation, or test dataset splits used for training these models.
Hardware Specification No The paper only mentions using a GPU device in a code snippet ('device="gpu:0"'), but it does not specify any particular GPU model, CPU, or other hardware specifications used for running the experiments or demonstrations.
Software Dependencies No The paper mentions "Paddle Paddle" as the deep learning framework for Interpret DL and other frameworks like PyTorch and TensorFlow for comparative libraries. It also mentions "Python Package Index (Py PI)" and "sphinx documentation generator". However, it does not provide specific version numbers for any of these software dependencies.
Experiment Setup No The paper demonstrates the Interpret DL toolkit by showing how to interpret a pre-trained ResNet-50 model and a sentiment classifier. It does not provide any specific details about the experimental setup for training these models, such as hyperparameters, optimization settings, or training schedules.