Added tests for wandblogger (fc242f71) · Commits · 钟慕尧 / deepchem

deepchem/models/init.py

+1 −0

Original line number	Diff line number	Diff line
		@@ -6,6 +6,7 @@ from deepchem.models.models import Model
		from deepchem.models.keras_model import KerasModel
		from deepchem.models.multitask import SingletaskToMultitask
		from deepchem.models.callbacks import ValidationCallback
		from deepchem.models.wandblogger import WandbLogger

		from deepchem.models.fcnet import MultitaskRegressor
		from deepchem.models.fcnet import MultitaskClassifier

deepchem/models/tests/test_kerasmodel.py

+63 −34

Original line number	Diff line number	Diff line
		import os
		import unittest
		import math
		import deepchem as dc
		import numpy as np
		import tensorflow as tf
		@@ -18,8 +18,7 @@ def test_overfit_graph_model():
		logits = tf.keras.layers.Dense(1)(hidden)
		outputs = tf.keras.layers.Activation('sigmoid')(logits)
		keras_model = tf.keras.Model(inputs=inputs, outputs=[outputs, logits])
		model = dc.models.KerasModel(
		keras_model,
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.SigmoidCrossEntropy(),
		output_types=['prediction', 'loss'],
		learning_rate=0.005)
		@@ -47,8 +46,9 @@ def test_overfit_sequential_model():
		tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1, activation='sigmoid')
		])
		model = dc.models.KerasModel(
		keras_model, dc.models.losses.BinaryCrossEntropy(), learning_rate=0.005)
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.BinaryCrossEntropy(),
		learning_rate=0.005)
		model.fit(dataset, nb_epoch=1000)
		prediction = np.squeeze(model.predict_on_batch(X))
		assert np.array_equal(y, np.round(prediction))
		@@ -69,8 +69,7 @@ def test_fit_use_all_losses():
		tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1, activation='sigmoid')
		])
		model = dc.models.KerasModel(
		keras_model,
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.BinaryCrossEntropy(),
		learning_rate=0.005,
		log_frequency=10)
		@@ -92,8 +91,9 @@ def test_fit_on_batch():
		tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1, activation='sigmoid')
		])
		model = dc.models.KerasModel(
		keras_model, dc.models.losses.BinaryCrossEntropy(), learning_rate=0.005)
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.BinaryCrossEntropy(),
		learning_rate=0.005)
		i = 0
		for X, y, w, ids in dataset.iterbatches(model.batch_size, 500):
		i += 1
		@@ -113,8 +113,9 @@ def test_checkpointing():
		keras_model1 = tf.keras.Sequential([tf.keras.layers.Dense(10)])
		keras_model2 = tf.keras.Sequential([tf.keras.layers.Dense(10)])
		model1 = dc.models.KerasModel(keras_model1, dc.models.losses.L2Loss())
		model2 = dc.models.KerasModel(
		keras_model2, dc.models.losses.L2Loss(), model_dir=model1.model_dir)
		model2 = dc.models.KerasModel(keras_model2,
		dc.models.losses.L2Loss(),
		model_dir=model1.model_dir)

		# Check that they produce different results.

		@@ -148,8 +149,9 @@ def test_fit_restore():
		tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1, activation='sigmoid')
		])
		model = dc.models.KerasModel(
		keras_model, dc.models.losses.BinaryCrossEntropy(), learning_rate=0.005)
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.BinaryCrossEntropy(),
		learning_rate=0.005)
		model.fit(dataset, nb_epoch=1000)
		prediction = np.squeeze(model.predict_on_batch(X))
		assert np.array_equal(y, np.round(prediction))
		@@ -161,8 +163,7 @@ def test_fit_restore():
		tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1, activation='sigmoid')
		])
		model2 = dc.models.KerasModel(
		keras_model2,
		model2 = dc.models.KerasModel(keras_model2,
		dc.models.losses.BinaryCrossEntropy(),
		model_dir=model.model_dir)
		model2.fit(dataset, nb_epoch=1, restore=True)
		@@ -188,8 +189,8 @@ def test_uncertainty():
		output = tf.keras.layers.Dense(n_features)(dropout)
		log_var = tf.keras.layers.Dense(n_features)(dropout)
		var = tf.keras.layers.Activation(tf.exp)(log_var)
		keras_model = tf.keras.Model(
		inputs=[inputs, switch], outputs=[output, var, output, log_var])
		keras_model = tf.keras.Model(inputs=[inputs, switch],
		outputs=[output, var, output, log_var])

		def loss(outputs, labels, weights):
		diff = labels[0] - outputs[0]
		@@ -206,8 +207,8 @@ def test_uncertainty():
		deterministic=True,
		pad_batches=True):
		for epoch in range(epochs):
		for (X_b, y_b, w_b, ids_b) in dataset.iterbatches(
		batch_size=self.batch_size,
		for (X_b, y_b, w_b,
		ids_b) in dataset.iterbatches(batch_size=self.batch_size,
		deterministic=deterministic,
		pad_batches=pad_batches):
		if mode == 'predict':
		@@ -283,8 +284,7 @@ def test_tensorboard():
		keras_model = tf.keras.Sequential([
		tf.keras.layers.Dense(2, activation='softmax'),
		])
		model = dc.models.KerasModel(
		keras_model,
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.CategoricalCrossEntropy(),
		tensorboard=True,
		log_frequency=1)
		@@ -297,6 +297,35 @@ def test_tensorboard():
		assert file_size > 0


		def test_wandblogger():
		"""Test logging to Weights & Biases."""
		# Load dataset and Models
		tasks, datasets, transformers = dc.molnet.load_delaney(featurizer='ECFP',
		splitter='random')
		train_dataset, valid_dataset, test_dataset = datasets
		keras_model = tf.keras.Sequential(
		[tf.keras.layers.Dense(10, activation='relu'),
		tf.keras.layers.Dense(1)])
		metric = dc.metrics.Metric(dc.metrics.pearson_r2_score)
		wandblogger = dc.models.WandbLogger(train_dataset=train_dataset,
		eval_dataset=valid_dataset,
		metrics=[metric],
		logging_strategy="step",
		anonymous="allow",
		save_run_history=True)
		model = dc.models.KerasModel(keras_model,
		dc.models.losses.L2Loss(),
		wandb_logger=wandblogger)
		model.fit(train_dataset, nb_epoch=10)

		run_data = wandblogger.run_history._data
		valid_score = model.evaluate(valid_dataset, [metric], transformers)

		assert math.isclose(valid_score["pearson_r2_score"],
		run_data['eval/pearson_r2_score'],
		abs_tol=0.0005)


		def test_fit_variables():
		"""Test training a subset of the variables in a model."""

deepchem/models/wandblogger.py

+9 −0

Original line number	Diff line number	Diff line
		import math
		import copy
		import logging
		import importlib.util
		from typing import List, Optional, Union
		@@ -8,6 +9,7 @@ from deepchem.models.callbacks import ValidationCallback

		logger = logging.getLogger(__name__)


		def is_wandb_available():
		return importlib.util.find_spec("wandb") is not None

		@@ -40,6 +42,7 @@ class WandbLogger(object):
		anonymous: Optional[str] = "never",
		log_model: Optional[bool] = False,
		log_dataset: Optional[bool] = False,
		save_run_history: Optional[bool] = False,
		**kwargs):
		"""Parameters
		----------
		@@ -71,6 +74,8 @@ class WandbLogger(object):
		whether to log the model to W&B
		log_dataset: bool
		whether to log the dataset to W&B
		save_run_history: bool
		whether to save the run history to the logger at the end (for testing purposes)
		"""

		assert is_wandb_available(
		@@ -104,6 +109,7 @@ class WandbLogger(object):
		self.log_model = log_model
		self.log_dataset = log_dataset
		self.save_dir = save_dir
		self.save_run_history = save_run_history

		# set wandb init arguments
		self.wandb_init_params = dict(name=name,
		@@ -199,7 +205,10 @@ class WandbLogger(object):

		def finish(self):
		"""Finishes and closes the W&B run.
		Save run history data as field if configured to do that.
		"""
		if self.save_run_history:
		self.run_history = copy.deepcopy(self.wandb_run.history)
		self.wandb_run.finish()

		def update_config(self, config_data):

Admin message