On the Existence of Universal Lottery Tickets

Authors: Rebekka Burkholz, Nilanjana Laha, Rajarshi Mukherjee, Alkis Gotovos

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

Reproducibility Variable Result LLM Response
Research Type Experimental To showcase the practical relevance of our main theorems, we conduct two types of experiments on a machine with Intel(R) Core(TM) i9-10850K CPU @ 3.60GHz processor and GPU NVIDIA Ge Force RTX 3080 Ti.
Researcher Affiliation Academia Rebekka Burkholz CISPA Helmholtz Center for Information Security burkholz@cispa.de Nilanjana Laha, Rajarshi Mukherjee Harvard T.H. Chan School of Public Health rmukherj@hsph.harvard.edu Alkis Gotovos MIT CSAIL alkisg@mit.edu
Pseudocode No The paper describes methods and proofs but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks.
Open Source Code Yes Code for the experiments is publicly available in the Github repository Universal LT, which can be accessed with the following url https://github.com/Relational ML/ Universal LT.
Open Datasets Yes In the second type of experiments, we train our mother networks with edge-popup (Ramanujan et al., 2020) on MNIST (Le Cun & Cortes, 2010) for 100 epochs based on SGD with momentum 0.9, weight decay 0.0001, batch size 128, and target sparsity 0.5.
Dataset Splits No The paper mentions training on MNIST but does not specify how the dataset was split into training, validation, and test sets, or provide percentages/counts for these splits.
Hardware Specification Yes To showcase the practical relevance of our main theorems, we conduct two types of experiments on a machine with Intel(R) Core(TM) i9-10850K CPU @ 3.60GHz processor and GPU NVIDIA Ge Force RTX 3080 Ti.
Software Dependencies No The paper mentions training on MNIST with SGD and edge-popup but does not provide specific version numbers for any software libraries or dependencies (e.g., Python, PyTorch, TensorFlow, CUDA versions).
Experiment Setup Yes In the second type of experiments, we train our mother networks with edge-popup (Ramanujan et al., 2020) on MNIST (Le Cun & Cortes, 2010) for 100 epochs based on SGD with momentum 0.9, weight decay 0.0001, batch size 128, and target sparsity 0.5.