Preventing Posterior Collapse with delta-VAEs
Authors: Ali Razavi, Aaron van den Oord, Ben Poole, Oriol Vinyals
ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 4 EXPERIMENTS |
| Researcher Affiliation | Industry | Ali Razavi Deepmind alirazavi@google.com Aaron van den Oord Deepmind avdnoord@google.com Ben Poole Google Brain pooleb@google.com Oriol Vinyals Deepmind vinyals@google.com |
| Pseudocode | No | The paper describes methods in text and mathematical derivations but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide an unambiguous statement about releasing the source code for the described methodology or a direct link to a code repository. |
| Open Datasets | Yes | We applied our method to generative modeling of images on the CIFAR-10 (Krizhevsky et al.) and downsampled Image Net (Deng et al., 2009) (32 32 as prepared in van den Oord et al. (2016a)) datasets. ... For our experiments on natural language, we used the 1 Billion Words or LM1B (Chelba et al., 2013) dataset... |
| Dataset Splits | No | Table 1 includes 'Image Net 32 32 Valid' and Section 4.3 mentions 'CIFAR-10 training set' and 'heldout data'. While these imply the use of validation and training/test sets, the paper does not specify the exact split percentages, absolute sample counts, or explicitly reference predefined splits with citations for these datasets to allow full reproduction of the data partitioning. |
| Hardware Specification | Yes | Our experiments on natural images were conducted on Google Cloud TPU accelerators. For Image Net, we used 128 TPU cores with batch size of 1024. We used 8 TPU cores for CIFAR-10 with batch size of 64. |
| Software Dependencies | No | The paper mentions using 'TensorFlow (Abadi et al., 2016)' and 'Tensor2Tensor (Vaswani et al., 2018) codebase' but does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | The exact hyper-parameters of our network is detailed in Table 4. ... The exact hyper-parameters are summarized in Table 5. |