Parametric Visual Program Induction with Function Modularization

Authors: Xuguang Duan, Xin Wang, Ziwei Zhang, Wenwu Zhu

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the superiority of the proposed method on three visual program induction datasets involving parametric primitive functions. Experimental results show that our proposed model is able to significantly outperform the state-of-the-art baseline methods in terms of generating accurate programs.
Researcher Affiliation Academia 1Department of Computer Science and Technology, Tsinghua University, Beijing, China. Correspondence to: Xin Wang <xin wang@tsinghua.edu.cn>, Wenwu Zhu <wwzhu@tsinghua.edu.cn>.
Pseudocode Yes Algorithm 1 H2MCTS
Open Source Code No The paper does not provide a direct link to open-source code for the methodology or explicitly state that the code is publicly available.
Open Datasets Yes we adopt the LATEX 2D drawing dataset (Ellis et al., 2018). ... we adopt the 3D-Shape dataset (Tian et al., 2019)
Dataset Splits No The paper specifies training and testing sets but does not explicitly mention a separate validation set split or its details (e.g., percentages or counts).
Hardware Specification Yes The whole framework is launched on a GPU server with two Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz CPU processors and two Nvidia Ge Force RTX 3090 GPU processors.
Software Dependencies No The paper mentions using Python libraries like numpy and torch, and a distributed system based on Ray, but does not provide specific version numbers for these software components.
Experiment Setup No The paper mentions a scaling hyper-parameter β in its score function (Eq. 11) but does not provide its specific value. Other typical hyperparameters like learning rate, batch size, or epochs are not specified.