Efficient optimization of loops and limits with randomized telescoping sums

Authors: Alex Beatson, Ryan P Adams

ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our adaptive RT estimators on a range of applications including meta-optimization of learning rates, variational inference of ODE parameters, and training an LSTM to model long sequences. Section 7 presents experimental results.
Researcher Affiliation Academia 1Department of Computer Science, Princeton University, Princeton, NJ, USA.
Pseudocode Yes Appendix A presents algorithm pseudocode.
Open Source Code Yes Code may be found at https://github.com/PrincetonLIPS/randomized_telescopes.
Open Datasets Yes We next experiment with meta-optimization of a learning rate on MNIST. Finally, we study a high-dimensional optimization problem: training an LSTM to model sequences on enwik8.
Dataset Splits Yes There are 205 unique tokens. We use the first 90M, 5M, and 5M characters as the training, evaluation, and test sets.
Hardware Specification No The paper does not explicitly describe any specific hardware components (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies or libraries used in the experiments.
Experiment Setup Yes Lotka-Volterra ODE: 'batch size of 64... and a learning rate of 0.01. Evaluation is performed with a batch size of 512.' MNIST: 'Optimization is performed with a batch size of 100.' 'The outer optimization is performed with a learning rate of 0.01.' enwik8 LSTM: 'The optimization is performed with a learning rate of 2.2.' General: 'the tuning frequency K is set to 5, and the exponential moving average weight α is set to 0.9.'