Towards Croppable Implicit Neural Representations

Authors: Maor Ashkenazi, Eran Treister

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our method on various signal encoding tasks and compare it to alternatives.
Researcher Affiliation Academia Maor Ashkenazi Ben-Gurion University of the Negev maorash@post.bgu.ac.il Eran Treister Ben-Gurion University of the Negev erant@cs.bgu.ac.il
Pseudocode Yes Algorithm 1 Automatic Partitioning
Open Source Code Yes Code is available at https://github.com/maorash/Local-Global-INRs.
Open Datasets Yes Next, from the DIV2K dataset [2], we have randomly selected a subset of 25 images which were downsampled by a factor of four before training.
Dataset Splits No The paper discusses training and testing but does not explicitly describe the use of a validation set or its split ratios for hyperparameter tuning or model selection.
Hardware Specification Yes We ran the experiments multiple times on a single Nvidia RTX3090.
Software Dependencies No The paper mentions using 'Adam W scheduler [26]' and 'Adam optimizer [22]', but does not specify version numbers for general software dependencies or libraries (e.g., PyTorch version, Python version).
Experiment Setup Yes The full configuration and networks size for all experiments is in Appendix B.