Experiments/kyle models

[kyle-woodward]

New:

• added mobilenet_v3small and vgg16 model fns to models.py
• model_fit.py is now a CLI that takes -c --config file (.yml)

Notes:
I figured if we name our runc.yml files uniquely with our intials then its fine to push them and we won't overwrite each others, and we can test each other's out.

Upon running model_fit.py wiht vgg16 as the model, I get an error in the confusion_matrix callback from sk-learn. it does not raise at first iteration of training though. Could be a parameteraization of sklearn.metrics.confusion_matrix to ignore inconsistent per-label sample sizes??

Epoch 1/100
4/4 [==============================] - 1s 106ms/stepe_coef: 0.6609 - binary_accuracy: 0.6362
Traceback (most recent call last):
  File "C:\fao-models\fao_models\model_fit.py", line 139, in <module>
    main()
  File "C:\fao-models\fao_models\model_fit.py", line 127, in main
    history = model.fit(
  File "C:\Users\kyle\anaconda3\envs\sigma\lib\site-packages\keras\utils\traceback_utils.py", line 70, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "C:\fao-models\fao_models\models.py", line 71, in on_epoch_end
    self.log_confusion_matrix(epoch=epoch)
  File "C:\fao-models\fao_models\models.py", line 79, in log_confusion_matrix
    cm = confusion_matrix(self.test_labels, test_pred)
  File "C:\Users\kyle\anaconda3\envs\sigma\lib\site-packages\sklearn\utils\_param_validation.py", line 211, in wrapper
    return func(*args, **kwargs)
  File "C:\Users\kyle\anaconda3\envs\sigma\lib\site-packages\sklearn\metrics\_classification.py", line 326, in confusion_matrix
    y_type, y_true, y_pred = _check_targets(y_true, y_pred)
  File "C:\Users\kyle\anaconda3\envs\sigma\lib\site-packages\sklearn\metrics\_classification.py", line 84, in _check_targets
    check_consistent_length(y_true, y_pred)
  File "C:\Users\kyle\anaconda3\envs\sigma\lib\site-packages\sklearn\utils\validation.py", line 409, in check_consistent_length
    raise ValueError(
ValueError: Found input variables with inconsistent numbers of samples: [97, 128]

[jdilger]

LGTM 😄