Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Decentralized Local Stochastic Extra-Gradient for Variational Inequalities
Authors: Aleksandr Beznosikov, Pavel Dvurechenskii, Anastasiia Koloskova, Valentin Samokhin, Sebastian U. Stich, Alexander Gasnikov
NeurIPS 2022 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We verify our theoretical results in numerical experiments and demonstrate the practical effectiveness of the proposed scheme. In particular, we train the DCGAN [79] architecture on the CIFAR-10 [51] dataset. and 5 Experiments In this section, we present two sets of experiments to validate the performance of Algorithm 1. |
| Researcher Affiliation | Collaboration | Aleksandr Beznosikov Innopolis University , MIPT , HSE University and Yandex |
| Pseudocode | Yes | Algorithm 1 Extra Step Time-Varying Gossip Method |
| Open Source Code | No | The paper includes a checklist entry stating 'Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes]', but does not provide a direct URL or explicitly state in the main text that the code is in the supplementary material. |
| Open Datasets | Yes | We consider the CIFAR-10 [51] dataset containing 60000 images, equally distributed over 10 classes. |
| Dataset Splits | No | The paper mentions partitioning the CIFAR-10 dataset into 16 subsets for nodes and states 'equally distributed over 10 classes', but it does not provide specific training, validation, or test splits (e.g., percentages or counts) or details on how these splits were handled for reproduction. |
| Hardware Specification | No | We simulate a distributed setup of 16 nodes on two GPUs and use Ray [67]. This mentions 'two GPUs' but does not specify the exact model or type of GPUs used. |
| Software Dependencies | No | The paper mentions using 'Ray [67]' and 'Adam [42] as the optimizer', but it does not provide specific version numbers for these or any other key software components used in the experiments. |
| Experiment Setup | Yes | We use the same learning rate equal to 0.002 for the generator and discriminator. The rest of the parameters and features of the architecture can be found in the supplementary material. |