tag:github.com,2008:https://github.com/cedrickchee/capsule-net-pytorch/releases 2018-01-30T15:05:22Z tag:github.com,2008:Repository/109484232/v0.4.0 2018-01-30T15:20:12Z <p>Version 0.4.0 is the first stable release as we have reached 100% completion state based on what we have planned out to do. The details of the changes are in the <a href="/cedrickchee/capsule-net-pytorch/blob/v0.4.0/CHANGELOG.md">CHANGELOG</a>. Please take a look.</p> <p>Thank you and happy hacking!</p> cedrickchee tag:github.com,2008:Repository/109484232/v0.3.0 2017-11-26T18:53:09Z <p>refactor decoder network</p> cedrickchee tag:github.com,2008:Repository/109484232/v0.2.0 2017-11-26T10:12:22Z <p>enhancement</p> cedrickchee tag:github.com,2008:Repository/109484232/v0.1.0 2017-11-12T16:24:17Z <h1>Added</h1> <ul> <li>Implemented reconstruction loss.</li> <li>Saving reconstructed image as file.</li> <li>Improve training speed by using PyTorch DataParallel to wrap our model. <ul> <li>PyTorch will parallelized the model and data over multiple GPUs.</li> </ul> </li> <li>Supports training: <ul> <li>on CPU (tested with macOS Sierra)</li> <li>on one GPU (tested with 1 Tesla K80 GPU)</li> <li>on multiple GPU (tested with 8 GPUs)</li> <li>with or without CUDA (tested with CUDA version 8.0.61)</li> <li>cuDNN 5 (tested with cuDNN 5.1.3)</li> </ul> </li> </ul> <h1>Changed</h1> <ul> <li>More intuitive variable naming.</li> </ul> <h1>Fixed</h1> <ul> <li>Resolve Pylint warnings and reformat code.</li> <li>Missing square in equation 4 for margin (class) loss.</li> </ul> cedrickchee tag:github.com,2008:Repository/109484232/v0.0.1 2017-11-06T02:06:16Z <p>Initial release. API is stable. The code runs. It's safe to use for development but not ready for general production usage.</p> cedrickchee