Deepml

- Helps you avoid lots of boilerplate code you write every time from scratch while training a model in pytorch.
- Easy to use deepml API.
- Supports Semantic segmentation.
- You can also define your custom machine learning task. deepml library is extensible.
- With 3-5 lines of deepml API you can start training your model in pytorch:

from deepml.tasks import ImageClassification
from deepml.train import Learner
task = ImageClassification(model, 'experiment1')
learner = Learner(ml_task, optimizer, criterion)
learner.fit(train_loader, val_loader)


What's Changed
* Feature/multi predictor by sagar-rathod in https://github.com/sagar-rathod/deep-ml/pull/11
* Feature/multi predictor by sagar-rathod in https://github.com/sagar-rathod/deep-ml/pull/12
* Update notebook links in README by sagar-rathod in https://github.com/sagar-rathod/deep-ml/pull/13


**Full Changelog**: https://github.com/sagar-rathod/deep-ml/compare/v1.1.0...v2.0.0

1. Adds **SimplePredictor** class which can be used for any custom data model training. Currently there is no inherent support
for video classification in the library however, you can you SimplePredictor class which will avoid writing to tensorboard.

2. Adds new classification metric **Matthews correlation coefficient**.

3. Supports learning rate scheduler policy (batch or epoch).

4. Use tensorboard to track learning rate for different parameter groups used in torch optimizer.

1. Supports image classification/regression using CNN.
2. Quickly visualize model's prediction.
3. Use tensorboard to track metrics and predictions while model trains.