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].
Adaptive Antithetic Sampling for Variance Reduction
Authors: Hongyu Ren, Shengjia Zhao, Stefano Ermon
ICML 2019 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We test our adaptive Gaussian antithetic on three tasks. The ๏ฌrst one is a controlled synthetic task, where we verify that our model demonstrates expected behavior. The second task is Bayesian inference, where we reduce variance and achieve better estimation of the posterior. The third task is improving stochastic gradient descent training of generative adversarial networks, where we reduce variance and obtain quantitative improvements in terms of inception scores for the same wall-clock time. |
| Researcher Affiliation | Academia | Hongyu Ren * 1 Shengjia Zhao * 1 Stefano Ermon 1 1Stanford University. |
| Pseudocode | No | The information is insufficient. The paper describes its methods and procedures using prose and mathematical equations but does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The information is insufficient. The paper does not contain any statements about making code open source or providing links to a code repository. |
| Open Datasets | Yes | We ๏ฌrst train a variational autoencoder on MNIST and Omniglot for the pretrained model q(z|x), p(x|z). and We train WGAN-GP (Gulrajani et al., 2017) on MNIST dataset and Fashion MNIST. |
| Dataset Splits | No | The information is insufficient. While the paper mentions training and testing on datasets like MNIST and Omniglot, it does not provide specific percentages or sample counts for training, validation, or test splits to reproduce the data partitioning. |
| Hardware Specification | No | The information is insufficient. The paper does not specify any particular hardware (e.g., CPU, GPU models, or cloud computing instances with their specifications) used for running the experiments. |
| Software Dependencies | No | The information is insufficient. The paper does not provide specific names or version numbers for any ancillary software dependencies (e.g., libraries, frameworks, or solvers). |
| Experiment Setup | No | The information is insufficient. The paper discusses general experiment applications but does not provide specific hyperparameter values (e.g., learning rate, number of epochs, optimizer details) or detailed system-level training settings required for reproduction. |