metadata
language: en
tags:
- pythae
- reproducibility
license: apache-2.0
This model was trained with pythae. It can be downloaded or reloaded using the method load_from_hf_hub
>>> from pythae.models import AutoModel
>>> model = AutoModel.load_from_hf_hub(hf_hub_path="clementchadebec/reproduced_iwae")
Reproducibility
This trained model reproduces the results of Table 1 in [1].
| Model | Dataset | Metric | Obtained value | Reference value |
|---|---|---|---|---|
| IWAE (n_samples=5) | Binary MNIST | NLL (5000 IS) | 87.85 (0.01) | 87.6 |
| IWAE (n_samples=50) | Binary MNIST | NLL (5000 IS) | 86.82 (0.01) | 87.1 |
[1] Burda, Y. et al, Importance Weighted Autoencoders, ArXiv:1509.00519