- 23 Oct, 2020 1 commit
-
-
F-G Fernandez authored
* fix: Fixed constructor typing in models._utils * fix: Fixed constructor typing in models.alexnet * fix: Fixed constructor typing in models.mnasnet * fix: Fixed constructor typing in models.squeezenet
-
- 22 Oct, 2020 1 commit
-
-
F-G Fernandez authored
* style: Added annotation typing for mnasnet * refactor: Removed un-necessary import
-
- 06 Apr, 2020 1 commit
-
-
Yonghye Kwon authored
cleanup
-
- 23 Sep, 2019 1 commit
-
-
Dmitry Belenko authored
* Add initial mnasnet impl * Remove all type hints, comply with PyTorch overall style * Expose models * Remove avgpool from features() and add separately * Fix python3-only stuff, replace subclasses with functions * fix __all__ * Fix typo * Remove conditional dropout * Make dropout functional * Addressing @fmassa's feedback, round 1 * Replaced adaptive avgpool with mean on H and W to prevent collapsing the batch dimension * Partially address feedback * YAPF * Removed redundant class vars * Update urls to releases * Add information to models.rst * Replace init with kaiming_normal_ in fan-out mode * Use load_state_dict_from_url * Fix depth scaling on first 2 layers * Restore initialization * Match reference implementation initialization for dense layer * Meant to use Kaiming * Remove spurious relu * Point to the newest 0.5 checkpoint * Latest pretrained checkpoint * Restore 1.0 checkpoint * YAPF * Implement backwards compat as suggested by Soumith * Update checkpoint URL * Move warnings up * Record a couple more function parameters * Update comment * Set the correct version such that if the BC-patched model is saved, it could be reloaded with BC patching again * Set a member var, not class var * Update mnasnet.py Remove unused member var as per review. * Update the path to weights
-
- 05 Jul, 2019 1 commit
-
-
ekka authored
This PR updates the docs of MnasNet
-
- 24 Jun, 2019 2 commits
-
-
Francisco Massa authored
-
Dmitry Belenko authored
* Add initial mnasnet impl * Remove all type hints, comply with PyTorch overall style * Expose models * Remove avgpool from features() and add separately * Fix python3-only stuff, replace subclasses with functions * fix __all__ * Fix typo * Remove conditional dropout * Make dropout functional * Addressing @fmassa's feedback, round 1 * Replaced adaptive avgpool with mean on H and W to prevent collapsing the batch dimension * Partially address feedback * YAPF * Removed redundant class vars * Update urls to releases * Add information to models.rst * Replace init with kaiming_normal_ in fan-out mode * Use load_state_dict_from_url
-
