Fix numerical problems with gradient by adding a very small delta (be07198b) · Commits · 钟慕尧 / deepchem

deepchem/models/tensorgraph/layers.py

+5 −2

Original line number	Diff line number	Diff line
		@@ -3294,9 +3294,12 @@ class ANIFeat(Layer):
		tensor2 = tf.stack([coordinates] * max_atoms, axis=2)

		# Calculate pairwise distance
		# d = tf.sqrt(
		# tf.nn.relu(
		# tf.reduce_sum(tf.squared_difference(tensor1, tensor2), axis=3)))
		d = tf.sqrt(
		tf.nn.relu(
		tf.reduce_sum(tf.squared_difference(tensor1, tensor2), axis=3)))
		tf.reduce_sum(tf.squared_difference(tensor1, tensor2), axis=3)+1e-7)
		# d = tf.reduce_sum(tf.squared_difference(tensor1, tensor2), axis=3)
		# Masking for valid atom index
		d = d * flags
		return d

+19 −9

Original line number	Diff line number	Diff line
		@@ -13,14 +13,24 @@ class TestANIRegression(unittest.TestCase):

		def setUp(self):

		max_atoms = 3

		X = np.array([[1, 5.0, 3.2, 1.1], [6, 1.0, 3.4, -1.1], [1, 2.3, 3.4, 2.2]])

		X = X.reshape((1, X.shape[0], X.shape[1]))
		max_atoms = 4

		X = np.array([
		[
		[1, 5.0, 3.2, 1.1],
		[6, 1.0, 3.4, -1.1],
		[1, 2.3, 3.4, 2.2],
		[0, 0, 0, 0],
		],
		[
		[8, 2.0, -1.4, -1.1],
		[7, 6.3, 2.4, 3.2],
		[0, 0, 0, 0],
		[0, 0, 0, 0],
		]
		])

		y = np.array([2.0])
		y = y.reshape((1, 1))
		y = np.array([2.0, 1.1])

		layer_structures = [128, 128, 64]
		atom_number_cases = [1, 6, 7, 8]
		@@ -32,7 +42,7 @@ class TestANIRegression(unittest.TestCase):
		"max_atoms": max_atoms,
		"layer_structures": layer_structures,
		"atom_number_cases": atom_number_cases,
		"batch_size": 1,
		"batch_size": 2,
		"learning_rate": 0.001,
		"use_queue": False,
		"mode": "regression",

+4 −4

Original line number	Diff line number	Diff line
		@@ -60,10 +60,10 @@ def load_roiterberg_ANI(mode="atomization"):
		hdf5files = [
		'ani_gdb_s01.h5',
		'ani_gdb_s02.h5',
		'ani_gdb_s03.h5',
		'ani_gdb_s04.h5',
		'ani_gdb_s05.h5',
		'ani_gdb_s06.h5',
		# 'ani_gdb_s03.h5',
		# 'ani_gdb_s04.h5',
		# 'ani_gdb_s05.h5',
		# 'ani_gdb_s06.h5',
		# 'ani_gdb_s07.h5',
		# 'ani_gdb_s08.h5'
		]