Dr Ed Morris comments

Results 27 comments of


                                            Dr Ed Morris

EARLY_STOPPING_PATIENCE config keyword sitting in wrong place in config.py

Also, DATA.TRANSFORMS.PARAMS.AGGRESIVE is spelt incorrectly. It should be AGGRESSIVE!

ModuleNotFoundError: No module named 'pytorchcv.utils'

Thanks for this. I missed this dependency from the list. Will update the relevant files.

Add inference methods to Trainer class.

As part of commit 008c52961434708a5cb7d1fb29cc2d08eb8268ac the method `update_model_from_checkpoint` was added to the `IgniteTrainer `class. This allows for the loading of an Ignite checkpoint file, which contains model weights, either of...

Modify setup.py so that it has entry points for CLI application added.

setup.py file allows for the use of CLI apps as a way of entry into a python application. Instead of having invoke the python interpreter on the command line, e.g....

Integrate torchmetrics package for all metrics used.

Use package [TorchMetrics](https://github.com/PyTorchLightning/metrics) for all metrics.

[Core] Submitted containerized job is stuck in pending mode

Still blocked by #42058 as well. @salotz just so you're aware, there is PR #42121 awaiting review which will solve this issue.

Compilation Fails on AARCH64 / ARM64 Due to x86 Specific Compiler Flags

Suggest the following addition to the `CMakeLists.txt` to accommodate: Currently the compiler flag options cover the following: ``` if(MSVC) SET(CMAKE_CXX_FLAGS "/EHsc /arch:AVX2") SET(CMAKE_CXX_FLAGS_DEBUG "/W4") elseif(DARWIN) SET(CMAKE_CXX_FLAGS "-Wno-int-in-bool-context -std=c++11") SET(CMAKE_CXX_FLAGS_DEBUG "-Wall")...