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 Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Spread Divergence
Authors: Mingtian Zhang, Peter Hayes, Thomas Bird, Raza Habib, David Barber
ICML 2020 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show results for a toy experiment in figure(4), learning the mixing matrix A in a deterministic non-square setting. MNIST Experiment: We trained a δ-VAE on MNIST (Le Cun et al., 2010)... Celeb A Experiment: We trained a δ-VAE on the Celeb A dataset (Liu et al., 2015)... Table 1. Celeb A FID Scores. |
| Researcher Affiliation | Academia | Mingtian Zhang 1 Peter Hayes 1 Tom Bird 1 Raza Habib 1 David Barber 1 1Department of Computer Science, University College London, UK. Correspondence to: Mingtian Zhang <EMAIL>. |
| Pseudocode | No | The paper describes its methods using prose and mathematical equations but does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper mentions a third-party tool's GitHub link ('github.com/bioinf-jku/TTUR') but does not provide any concrete access to the source code for the methodology described in this paper. |
| Open Datasets | Yes | MNIST Experiment: We trained a δ-VAE on MNIST (Le Cun et al., 2010)... and Celeb A Experiment: We trained a δ-VAE on the Celeb A dataset (Liu et al., 2015)... |
| Dataset Splits | No | The paper mentions using 'training data' for experiments on MNIST and Celeb A datasets but does not provide specific dataset split information (exact percentages, sample counts, or detailed splitting methodology) for reproduction. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers, needed to replicate the experiment. |
| Experiment Setup | Yes | We first train with one epoch a standard VAE as initialization to all models and keep the latent code z ∼ N (z 0, IZ) fixed when sampling from these models thereafter, so we can more easily compare the sample quality. gθ( ) is a neural network that contains 3 feed forward layers. (MNIST) gθ( ) is a neural network contains 4 convolution layers. (Celeb A) We use Sy=1, Sz=1000 samples and 2000 EM iterations to estimate A. (ICA). |