Merge pull request #1954 from borisdayma/feat_wandb

- [RDKit](http://www.rdkit.org/docs/Install.html)

- [simdna](https://github.com/kundajelab/simdna)

- [XGBoost](https://xgboost.readthedocs.io/en/latest/)

- [Weights & Biases](https://docs.wandb.com/)

- [Tensorflow Probability](https://www.tensorflow.org/probability)

## Installation

import tensorflow as tf

import sys

from deepchem.models.keras_model import is_wandb_available

if is_wandb_available():

  import wandb

class ValidationCallback(object):

  """Performs validation while training a KerasModel.

    if model.tensorboard:

      for key in scores:

        model._log_value_to_tensorboard(tag=key, simple_value=scores[key])

    if model.wandb:

      wandb.log(scores, step=step)

    if self.save_dir is not None:

      score = scores[self.metrics[self.save_metric].name]

      if not self.save_on_minimum:

from deepchem.trans import undo_transforms

from deepchem.utils.evaluate import GeneratorEvaluator

try:

  import wandb

  wandb.ensure_configured()

  if wandb.api.api_key is None:

    _has_wandb = False

    wandb.termwarn(

        "W&B installed but not logged in.  Run `wandb login` or set the WANDB_API_KEY env variable."

    )

  else:

    _has_wandb = True

except (ImportError, AttributeError):

  _has_wandb = False

def is_wandb_available():

  return _has_wandb

class KerasModel(Model):

  """This is a DeepChem model implemented by a Keras model.

               learning_rate=0.001,

               optimizer=None,

               tensorboard=False,

               wandb=False,

               log_frequency=100,

               **kwargs):

    """Create a new KerasModel.

      ignored.

    tensorboard: bool

      whether to log progress to TensorBoard during training

    wandb: bool

      whether to log progress to Weights & Biases during training

    log_frequency: int

      The frequency at which to log data. Data is logged using

      `logging` by default. If `tensorboard` is set, data is also

    else:

      self.optimizer = optimizer

    self.tensorboard = tensorboard

    # W&B logging

    if wandb and not is_wandb_available():

      logger.warning(

          "You set wandb to True but W&B is not installed. To use wandb logging, "

          "run `pip install wandb; wandb login` see https://docs.wandb.com/huggingface."

      )

    self.wandb = wandb and is_wandb_available()

    # Backwards compatibility

    if "tensorboard_log_frequency" in kwargs:

      logger.warning(

      if self.tensorboard and should_log:

        with self._summary_writer.as_default():

          tf.summary.scalar('loss', batch_loss, current_step)

      if self.wandb and should_log:

        wandb.log({'loss': batch_loss}, step=current_step)

    # Report final results.

    if averaged_batches > 0:

----------

DeepChem extensively uses `Keras`_ to build powerful machine learning models.

Training loss and validation metrics can be automatically logged to `Weights & Biases`_ with the following commands::

  # Install wandb in shell

  pip install wandb

  # Login in shell (required only once)

  wandb login

  # Start a W&B run in your script (refer to docs for optional parameters)

  wandb.init(project="my project")

  # Set `wandb` arg when creating `KerasModel`

  model = KerasModel(…, wandb=True)

.. _`Keras`: https://keras.io/

.. _`Weights & Biases`: http://docs.wandb.com/

.. autoclass:: deepchem.models.KerasModel

  :members: