torchdistill >= v1.0.0 support and updated documentation

Breaking changes

For torchdistill >= v1.0.0 support, most of the old scripts and configurations do not work with sc2bench v0.1.0.
If you want to use the old version, you can find the old scripts, configurations specifically with sc2bench < v.0.1.0, old dependencies, etc under legacy/

• Support torchdistill >= v1.0.0 (PR #145)

Model migration from torchdistill

Split computing models in torchdistill were migrated to sc2bench and available in PyTorch Hub.

• Add default short module name set (PR #158)
• Add custom object detection models in torchdistill (PR #156)
• Fix bugs (PR #155)
• Add skips_dropout (PR #154)
• Fix typos (PR #153)
• Forgot to add kwargs (PR #152)
• Fix incompatiblity (PR #151)
• Make it consistent with torchdistill ver (PR #150)
• Fix bugs in hubconf.py (PR #149)

Documentation

• Add Google Analytics ID (PR #147)
• Fix a typo (PR #146)

Misc.

• Update version (PR #160)
• 1 indentation is missed (PR #159)
• Fix anchors and references (PR #157)
• Update paths (PR #148)
• Update image links (PR #140)

Documentation

Documentation

• Fix typos (PR #138)
• Update documentation (PR #137)
• Fix a typo (PR #136)
• Change depth (PR #134)
• Add docstrings to loss (PR #130)
• Add docstrings to analysis (PR #129)
• Add docstrings to transforms package (PR #128)
• Make minor updates (PR #127)
• Make minor updates in docstrings (PR #126)
• Add docstrings to segmentation package (PR #124)
• Add docstrings to detection package (PR #123)

Misc.

• Update setup (PR #135)
• Fix a typo in link (PR #132)
• Update module names in configs (PR #131)
• Add a citation file (PR #125)
• Update checkpoint link (PR #122)
• Update publication (PR #121)
• Resolve name conflict (PR #120)
• Fix a typo (PR #119)
• Update versions (PRs #118, #139)

Workflows

• Add a GitHub Action for deploying Sphinx documentation (PR #133)

A set of minor updates

Bug fix

• Fix a bug regarding default args for analyze_model_size (PR #114)

New feature

• Add option to load ckpt for original model before splittable model instantiation (PR #115)

Misc.

• Update versions (PRs #98, #99, #116, #118)
• Update README (PR #100)
• Add a badge (PR #105)
• Add an issue template (PR #108)
• Remove apex and apex configs (PRs #110, #111)
• Update minimum requirements (PR #113)
• Add content, following the new template (PR #117)

Support the latest version of PyTorch

v0.0.1 has a potential bug when using the latest version of PyTorch (torch: 1.11.0, torchvision: 0.12.0).
This version uses the update package path for segmentation model in torchvision.

The first release of sc2bench package

The first release of sc2bench package with code for "SC2: Supervised Compression for Split Computing"