def _run_test_train_full_low_rank_wals(self, use_factors_weights_cache):
rows = 15
cols = 11
dims = 3
with ops.Graph().as_default(), self.cached_session():
data = np.dot(np.random.rand(rows, 3), np.random.rand(
3, cols)).astype(np.float32) / 3.0
indices = [[i, j] for i in xrange(rows) for j in xrange(cols)]
values = data.reshape(-1)
inp = sparse_tensor.SparseTensor(indices, values, [rows, cols])
model = factorization_ops.WALSModel(
rows,
cols,
dims,
regularization=1e-5,
row_weights=0,
col_weights=[0] * cols,
use_factors_weights_cache=use_factors_weights_cache)
self.simple_train(model, inp, 25)
row_factor = model.row_factors[0].eval()
col_factor = model.col_factors[0].eval()
self.assertAllClose(
data,
np.dot(row_factor, np.transpose(col_factor)),
rtol=0.01,
atol=0.01)
def _run_test_sum_weights(self, test_rows):
# test_rows: True to test row weights, False to test column weights.
num_rows = 5
num_cols = 5
unobserved_weight = 0.1
row_weights = [[8., 18., 28., 38., 48.]]
col_weights = [[90., 91., 92., 93., 94.]]
sparse_indices = [[0, 1], [2, 3], [4, 1]]
sparse_values = [666., 777., 888.]
unobserved = unobserved_weight * num_rows * num_cols
observed = 8. * 91. + 28. * 93. + 48. * 91.
# sparse_indices has three unique rows and two unique columns
observed *= num_rows / 3. if test_rows else num_cols / 2.
want_weight_sum = unobserved + observed
with ops.Graph().as_default(), self.cached_session() as sess:
wals_model = factorization_ops.WALSModel(
input_rows=num_rows,
input_cols=num_cols,
n_components=5,
unobserved_weight=unobserved_weight,
row_weights=row_weights,
col_weights=col_weights,
use_factors_weights_cache=False)
wals_model.initialize_op.run()
wals_model.worker_init.run()
update_factors = (wals_model.update_row_factors
if test_rows else wals_model.update_col_factors)
(_, _, _, _, sum_weights) = update_factors(
sp_input=sparse_tensor.SparseTensor(
indices=sparse_indices,
values=sparse_values,
dense_shape=[num_rows, num_cols]),
transpose_input=False)
got_weight_sum = sess.run(sum_weights)
self.assertNear(
got_weight_sum,
want_weight_sum,
err=.001,
msg="got weight sum [{}], want weight sum [{}]".format(
got_weight_sum, want_weight_sum))
def _run_test_train_matrix_completion_wals(self, use_factors_weights_cache):
rows = 11
cols = 9
dims = 4
def keep_index(x):
return not (x[0] + x[1]) % 4
with ops.Graph().as_default(), self.cached_session():
row_wts = 0.1 + np.random.rand(rows)
col_wts = 0.1 + np.random.rand(cols)
data = np.dot(np.random.rand(rows, 3), np.random.rand(
3, cols)).astype(np.float32) / 3.0
indices = np.array(
list(
filter(keep_index,
[[i, j] for i in xrange(rows) for j in xrange(cols)])))
values = data[indices[:, 0], indices[:, 1]]
inp = sparse_tensor.SparseTensor(indices, values, [rows, cols])
model = factorization_ops.WALSModel(
rows,
cols,
dims,
unobserved_weight=0.01,
regularization=0.001,
row_weights=row_wts,
col_weights=col_wts,
use_factors_weights_cache=use_factors_weights_cache)
self.simple_train(model, inp, 25)
row_factor = model.row_factors[0].eval()
col_factor = model.col_factors[0].eval()
out = np.dot(row_factor, np.transpose(col_factor))
for i in xrange(rows):
for j in xrange(cols):
if keep_index([i, j]):
self.assertNear(
data[i][j], out[i][j], err=0.4, msg="%d, %d" % (i, j))
else:
self.assertNear(0, out[i][j], err=0.5, msg="%d, %d" % (i, j))
def _wals_factorization_model_function(features, labels, mode, params):
"""Model function for the WALSFactorization estimator.
Args:
features: Dictionary of features. See WALSMatrixFactorization.
labels: Must be None.
mode: A model_fn.ModeKeys object.
params: Dictionary of parameters containing arguments passed to the
WALSMatrixFactorization constructor.
Returns:
A ModelFnOps object.
Raises:
ValueError: If `mode` is not recognized.
"""
assert labels is None
use_factors_weights_cache = (
params["use_factors_weights_cache_for_training"]
and mode == model_fn.ModeKeys.TRAIN)
use_gramian_cache = (params["use_gramian_cache_for_training"]
and mode == model_fn.ModeKeys.TRAIN)
max_sweeps = params["max_sweeps"]
model = factorization_ops.WALSModel(
params["num_rows"],
params["num_cols"],
params["embedding_dimension"],
unobserved_weight=params["unobserved_weight"],
regularization=params["regularization_coeff"],
row_init=params["row_init"],
col_init=params["col_init"],
num_row_shards=params["num_row_shards"],
num_col_shards=params["num_col_shards"],
row_weights=params["row_weights"],
col_weights=params["col_weights"],
use_factors_weights_cache=use_factors_weights_cache,
use_gramian_cache=use_gramian_cache)
# Get input rows and cols. We either update rows or columns depending on
# the value of row_sweep, which is maintained using a session hook.
input_rows = features[WALSMatrixFactorization.INPUT_ROWS]
input_cols = features[WALSMatrixFactorization.INPUT_COLS]
# TRAIN mode:
if mode == model_fn.ModeKeys.TRAIN:
# Training consists of the following ops (controlled using a SweepHook).
# Before a row sweep:
# row_update_prep_gramian_op
# initialize_row_update_op
# During a row sweep:
# update_row_factors_op
# Before a col sweep:
# col_update_prep_gramian_op
# initialize_col_update_op
# During a col sweep:
# update_col_factors_op
is_row_sweep_var = variable_scope.variable(
True,
trainable=False,
name="is_row_sweep",
collections=[ops.GraphKeys.GLOBAL_VARIABLES])
is_sweep_done_var = variable_scope.variable(
False,
trainable=False,
name="is_sweep_done",
collections=[ops.GraphKeys.GLOBAL_VARIABLES])
completed_sweeps_var = variable_scope.variable(
0,
trainable=False,
name=WALSMatrixFactorization.COMPLETED_SWEEPS,
collections=[ops.GraphKeys.GLOBAL_VARIABLES])
loss_var = variable_scope.variable(
0.,
trainable=False,
name=WALSMatrixFactorization.LOSS,
collections=[ops.GraphKeys.GLOBAL_VARIABLES])
# The root weighted squared error =
# \\(\sqrt( \sum_{i,j} w_ij * (a_ij - r_ij)^2 / \sum_{i,j} w_ij )\\)
rwse_var = variable_scope.variable(
0.,
trainable=False,
name=WALSMatrixFactorization.RWSE,
collections=[ops.GraphKeys.GLOBAL_VARIABLES])
summary.scalar("loss", loss_var)
summary.scalar("root_weighted_squared_error", rwse_var)
summary.scalar("completed_sweeps", completed_sweeps_var)
def create_axis_ops(sp_input, num_items, update_fn, axis_name):
"""Creates book-keeping and training ops for a given axis.
Args:
sp_input: A SparseTensor corresponding to the row or column batch.
num_items: An integer, the total number of items of this axis.
update_fn: A function that takes one argument (`sp_input`), and that
returns a tuple of
* new_factors: A float Tensor of the factor values after update.
* update_op: a TensorFlow op which updates the factors.
* loss: A float Tensor, the unregularized loss.
* reg_loss: A float Tensor, the regularization loss.
* sum_weights: A float Tensor, the sum of factor weights.
axis_name: A string that specifies the name of the axis.
Returns:
A tuple consisting of:
* reset_processed_items_op: A TensorFlow op, to be run before the
beginning of any sweep. It marks all items as not-processed.
* axis_train_op: A Tensorflow op, to be run during this axis' sweeps.
"""
processed_items_init = array_ops.fill(dims=[num_items],
value=False)
with ops.colocate_with(processed_items_init):
processed_items = variable_scope.variable(
processed_items_init,
collections=[ops.GraphKeys.GLOBAL_VARIABLES],
trainable=False,
name="processed_" + axis_name)
_, update_op, loss, reg, sum_weights = update_fn(sp_input)
input_indices = sp_input.indices[:, 0]
with ops.control_dependencies([
update_op,
state_ops.assign(loss_var, loss + reg),
state_ops.assign(rwse_var,
math_ops.sqrt(loss / sum_weights))
]):
with ops.colocate_with(processed_items):
update_processed_items = state_ops.scatter_update(
processed_items,
input_indices,
array_ops.ones_like(input_indices, dtype=dtypes.bool),
name="update_processed_{}_indices".format(axis_name))
with ops.control_dependencies([update_processed_items]):
is_sweep_done = math_ops.reduce_all(processed_items)
axis_train_op = control_flow_ops.group(
state_ops.assign(is_sweep_done_var, is_sweep_done),
state_ops.assign_add(
completed_sweeps_var,
math_ops.cast(is_sweep_done, dtypes.int32)),
name="{}_sweep_train_op".format(axis_name))
return processed_items.initializer, axis_train_op
reset_processed_rows_op, row_train_op = create_axis_ops(
input_rows, params["num_rows"],
lambda x: model.update_row_factors(sp_input=x,
transpose_input=False), "rows")
reset_processed_cols_op, col_train_op = create_axis_ops(
input_cols, params["num_cols"],
lambda x: model.update_col_factors(sp_input=x,
transpose_input=True), "cols")
switch_op = control_flow_ops.group(state_ops.assign(
is_row_sweep_var, math_ops.logical_not(is_row_sweep_var)),
reset_processed_rows_op,
reset_processed_cols_op,
name="sweep_switch_op")
row_prep_ops = [
model.row_update_prep_gramian_op, model.initialize_row_update_op
]
col_prep_ops = [
model.col_update_prep_gramian_op, model.initialize_col_update_op
]
init_op = model.worker_init
sweep_hook = _SweepHook(is_row_sweep_var, is_sweep_done_var, init_op,
row_prep_ops, col_prep_ops, row_train_op,
col_train_op, switch_op)
global_step_hook = _IncrementGlobalStepHook()
training_hooks = [sweep_hook, global_step_hook]
if max_sweeps is not None:
training_hooks.append(_StopAtSweepHook(max_sweeps))
return model_fn.ModelFnOps(mode=model_fn.ModeKeys.TRAIN,
predictions={},
loss=loss_var,
eval_metric_ops={},
train_op=control_flow_ops.no_op(),
training_hooks=training_hooks)
# INFER mode
elif mode == model_fn.ModeKeys.INFER:
projection_weights = features.get(
WALSMatrixFactorization.PROJECTION_WEIGHTS)
def get_row_projection():
return model.project_row_factors(
sp_input=input_rows,
projection_weights=projection_weights,
transpose_input=False)
def get_col_projection():
return model.project_col_factors(
sp_input=input_cols,
projection_weights=projection_weights,
transpose_input=True)
predictions = {
WALSMatrixFactorization.PROJECTION_RESULT:
control_flow_ops.cond(
features[WALSMatrixFactorization.PROJECT_ROW],
get_row_projection, get_col_projection)
}
return model_fn.ModelFnOps(mode=model_fn.ModeKeys.INFER,
predictions=predictions,
loss=None,
eval_metric_ops={},
train_op=control_flow_ops.no_op(),
training_hooks=[])
# EVAL mode
elif mode == model_fn.ModeKeys.EVAL:
def get_row_loss():
_, _, loss, reg, _ = model.update_row_factors(
sp_input=input_rows, transpose_input=False)
return loss + reg
def get_col_loss():
_, _, loss, reg, _ = model.update_col_factors(sp_input=input_cols,
transpose_input=True)
return loss + reg
loss = control_flow_ops.cond(
features[WALSMatrixFactorization.PROJECT_ROW], get_row_loss,
get_col_loss)
return model_fn.ModelFnOps(mode=model_fn.ModeKeys.EVAL,
predictions={},
loss=loss,
eval_metric_ops={},
train_op=control_flow_ops.no_op(),
training_hooks=[])
else:
raise ValueError("mode=%s is not recognized." % str(mode))
def _run_test_als_transposed(self, use_factors_weights_cache):
with ops.Graph().as_default(), self.cached_session():
self._wals_inputs = self.sparse_input()
col_init = np.random.rand(7, 3)
als_model = factorization_ops.WALSModel(
5,
7,
3,
col_init=col_init,
row_weights=None,
col_weights=None,
use_factors_weights_cache=use_factors_weights_cache)
als_model.initialize_op.run()
als_model.worker_init.run()
wals_model = factorization_ops.WALSModel(
5,
7,
3,
col_init=col_init,
row_weights=[0] * 5,
col_weights=[0] * 7,
use_factors_weights_cache=use_factors_weights_cache)
wals_model.initialize_op.run()
wals_model.worker_init.run()
sp_feeder = array_ops.sparse_placeholder(dtypes.float32)
# Here test partial row update with identical inputs but with transposed
# input for als.
sp_r_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, [3, 1], transpose=True).eval()
sp_r = np_matrix_to_tf_sparse(INPUT_MATRIX, [3, 1]).eval()
feed_dict = {sp_feeder: sp_r_t}
als_model.row_update_prep_gramian_op.run()
als_model.initialize_row_update_op.run()
process_input_op = als_model.update_row_factors(
sp_input=sp_feeder, transpose_input=True)[1]
process_input_op.run(feed_dict=feed_dict)
# Only updated row 1 and row 3, so only compare these rows since others
# have randomly initialized values.
row_factors1 = [
als_model.row_factors[0].eval()[1], als_model.row_factors[0].eval()[3]
]
# Testing row projection. Projection weight doesn't matter in this case
# since the model is ALS special case. Note that the ordering of the
# returned results will be preserved as the input feature vectors
# ordering.
als_projected_row_factors1 = als_model.project_row_factors(
sp_input=sp_feeder, transpose_input=True).eval(feed_dict=feed_dict)
feed_dict = {sp_feeder: sp_r}
wals_model.row_update_prep_gramian_op.run()
wals_model.initialize_row_update_op.run()
process_input_op = wals_model.update_row_factors(sp_input=sp_feeder)[1]
process_input_op.run(feed_dict=feed_dict)
# Only updated row 1 and row 3, so only compare these rows since others
# have randomly initialized values.
row_factors2 = [
wals_model.row_factors[0].eval()[1],
wals_model.row_factors[0].eval()[3]
]
for r1, r2 in zip(row_factors1, row_factors2):
self.assertAllClose(r1, r2, atol=1e-3)
# Note that the ordering of the returned projection results is preserved
# as the input feature vectors ordering.
self.assertAllClose(
als_projected_row_factors1, [row_factors2[1], row_factors2[0]],
atol=1e-3)
def _run_test_als(self, use_factors_weights_cache):
with ops.Graph().as_default(), self.cached_session():
self._wals_inputs = self.sparse_input()
col_init = np.random.rand(7, 3)
als_model = factorization_ops.WALSModel(
5,
7,
3,
col_init=col_init,
row_weights=None,
col_weights=None,
use_factors_weights_cache=use_factors_weights_cache)
als_model.initialize_op.run()
als_model.worker_init.run()
als_model.row_update_prep_gramian_op.run()
als_model.initialize_row_update_op.run()
process_input_op = als_model.update_row_factors(self._wals_inputs)[1]
process_input_op.run()
row_factors1 = [x.eval() for x in als_model.row_factors]
# Testing row projection. Projection weight doesn't matter in this case
# since the model is ALS special case.
als_projected_row_factors1 = als_model.project_row_factors(
self._wals_inputs).eval()
wals_model = factorization_ops.WALSModel(
5,
7,
3,
col_init=col_init,
row_weights=0,
col_weights=0,
use_factors_weights_cache=use_factors_weights_cache)
wals_model.initialize_op.run()
wals_model.worker_init.run()
wals_model.row_update_prep_gramian_op.run()
wals_model.initialize_row_update_op.run()
process_input_op = wals_model.update_row_factors(self._wals_inputs)[1]
process_input_op.run()
row_factors2 = [x.eval() for x in wals_model.row_factors]
for r1, r2 in zip(row_factors1, row_factors2):
self.assertAllClose(r1, r2, atol=1e-3)
self.assertAllClose(
als_projected_row_factors1,
[row for shard in row_factors2 for row in shard],
atol=1e-3)
# Here we test partial column updates.
sp_c = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[2, 0], shuffle=True).eval()
sp_feeder = array_ops.sparse_placeholder(dtypes.float32)
feed_dict = {sp_feeder: sp_c}
als_model.col_update_prep_gramian_op.run()
als_model.initialize_col_update_op.run()
process_input_op = als_model.update_col_factors(sp_input=sp_feeder)[1]
process_input_op.run(feed_dict=feed_dict)
col_factors1 = [x.eval() for x in als_model.col_factors]
# Testing column projection. Projection weight doesn't matter in this case
# since the model is ALS special case.
als_projected_col_factors1 = als_model.project_col_factors(
np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[2, 0], shuffle=False)).eval()
feed_dict = {sp_feeder: sp_c}
wals_model.col_update_prep_gramian_op.run()
wals_model.initialize_col_update_op.run()
process_input_op = wals_model.update_col_factors(sp_input=sp_feeder)[1]
process_input_op.run(feed_dict=feed_dict)
col_factors2 = [x.eval() for x in wals_model.col_factors]
for c1, c2 in zip(col_factors1, col_factors2):
self.assertAllClose(c1, c2, rtol=5e-3, atol=1e-2)
self.assertAllClose(
als_projected_col_factors1, [col_factors2[0][2], col_factors2[0][0]],
atol=1e-2)
def _run_test_process_input_transposed(self,
use_factors_weights_cache,
compute_loss=False):
with ops.Graph().as_default(), self.cached_session() as sess:
self._wals_inputs = self.sparse_input()
sp_feeder = array_ops.sparse_placeholder(dtypes.float32)
num_rows = 5
num_cols = 7
factor_dim = 3
wals_model = factorization_ops.WALSModel(
num_rows,
num_cols,
factor_dim,
num_row_shards=2,
num_col_shards=3,
regularization=0.01,
unobserved_weight=0.1,
col_init=self.col_init,
row_weights=self.row_wts,
col_weights=self.col_wts,
use_factors_weights_cache=use_factors_weights_cache)
wals_model.initialize_op.run()
wals_model.worker_init.run()
# Split input into multiple SparseTensors with scattered rows.
# Here the inputs are transposed. But the same constraints as described in
# the previous non-transposed test case apply to these inputs (before they
# are transposed).
sp_r0_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, [0, 3], transpose=True).eval()
sp_r1_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, [4, 1], shuffle=True, transpose=True).eval()
sp_r2_t = np_matrix_to_tf_sparse(INPUT_MATRIX, [2], transpose=True).eval()
sp_r3_t = sp_r1_t
input_scattered_rows = [sp_r0_t, sp_r1_t, sp_r2_t, sp_r3_t]
input_scattered_rows_non_duplicate = [sp_r0_t, sp_r1_t, sp_r2_t]
# Test updating row factors.
# Here we feed in scattered rows of the input.
# Note that the needed suffix of placeholder are in the order of test
# case name lexicographical order and then in the line order of where
# they appear.
wals_model.row_update_prep_gramian_op.run()
wals_model.initialize_row_update_op.run()
(_, process_input_op, unregularized_loss, regularization,
_) = wals_model.update_row_factors(
sp_input=sp_feeder, transpose_input=True)
factor_loss = unregularized_loss + regularization
for inp in input_scattered_rows:
feed_dict = {sp_feeder: inp}
process_input_op.run(feed_dict=feed_dict)
row_factors = [x.eval() for x in wals_model.row_factors]
self.assertAllClose(row_factors[0], self._row_factors_0, atol=1e-3)
self.assertAllClose(row_factors[1], self._row_factors_1, atol=1e-3)
# Test row projection.
# Using the specified projection weights for the 2 row feature vectors.
# This is expected to reproduce the same row factors in the model as the
# weights and feature vectors are identical to that used in model
# training.
projected_rows = wals_model.project_row_factors(
sp_input=sp_feeder,
transpose_input=True,
projection_weights=[0.5, 0.2])
# Don't specify the projection weight, so 1.0 will be used. The feature
# weights will be those specified in model.
projected_rows_no_weights = wals_model.project_row_factors(
sp_input=sp_feeder, transpose_input=True)
feed_dict = {
sp_feeder:
np_matrix_to_tf_sparse(
INPUT_MATRIX, [4, 1], shuffle=False, transpose=True).eval()
}
self.assertAllClose(
projected_rows.eval(feed_dict=feed_dict),
[self._row_factors_1[1], self._row_factors_0[1]],
atol=1e-3)
self.assertAllClose(
projected_rows_no_weights.eval(feed_dict=feed_dict),
[[1.915879, 1.992677, 1.109057], [0.569082, 0.715088, 0.31777]],
atol=1e-3)
if compute_loss:
# Test loss computation after the row update
loss = sum(
sess.run(
factor_loss * self.count_cols(inp) / num_rows,
feed_dict={sp_feeder: inp})
for inp in input_scattered_rows_non_duplicate)
true_loss = self.calculate_loss_from_wals_model(wals_model,
self._wals_inputs)
self.assertNear(
loss,
true_loss,
err=.001,
msg="After row update, computed loss [{}] does not match"
" true loss [{}]".format(loss, true_loss))
# Split input into multiple SparseTensors with scattered columns.
# Here the inputs are transposed. But the same constraints as described in
# the previous non-transposed test case apply to these inputs (before they
# are transposed).
sp_c0_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[0, 1], transpose=True).eval()
sp_c1_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[4, 2], transpose=True).eval()
sp_c2_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[5], transpose=True, shuffle=True).eval()
sp_c3_t = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[3, 6], transpose=True).eval()
sp_c4_t = sp_c2_t
input_scattered_cols = [sp_c0_t, sp_c1_t, sp_c2_t, sp_c3_t, sp_c4_t]
input_scattered_cols_non_duplicate = [sp_c0_t, sp_c1_t, sp_c2_t, sp_c3_t]
# Test updating column factors.
# Here we feed in scattered columns of the input.
wals_model.col_update_prep_gramian_op.run()
wals_model.initialize_col_update_op.run()
(_, process_input_op, unregularized_loss, regularization,
_) = wals_model.update_col_factors(
sp_input=sp_feeder, transpose_input=True)
factor_loss = unregularized_loss + regularization
for inp in input_scattered_cols:
feed_dict = {sp_feeder: inp}
process_input_op.run(feed_dict=feed_dict)
col_factors = [x.eval() for x in wals_model.col_factors]
self.assertAllClose(col_factors[0], self._col_factors_0, atol=1e-3)
self.assertAllClose(col_factors[1], self._col_factors_1, atol=1e-3)
self.assertAllClose(col_factors[2], self._col_factors_2, atol=1e-3)
# Test column projection.
# Using the specified projection weights for the 2 column feature vectors.
# This is expected to reproduce the same column factors in the model as
# the weights and feature vectors are identical to that used in model
# training.
projected_cols = wals_model.project_col_factors(
sp_input=sp_feeder,
transpose_input=True,
projection_weights=[0.4, 0.7])
# Don't specify the projection weight, so 1.0 will be used. The feature
# weights will be those specified in model.
projected_cols_no_weights = wals_model.project_col_factors(
sp_input=sp_feeder, transpose_input=True)
feed_dict = {sp_feeder: sp_c3_t}
self.assertAllClose(
projected_cols.eval(feed_dict=feed_dict),
[self._col_factors_1[0], self._col_factors_2[1]],
atol=1e-3)
self.assertAllClose(
projected_cols_no_weights.eval(feed_dict=feed_dict),
[[3.585139, -0.487476, -3.852232], [0.557937, 1.813907, 1.331171]],
atol=1e-3)
if compute_loss:
# Test loss computation after the col update
loss = sum(
sess.run(
factor_loss * self.count_rows(inp) / num_cols,
feed_dict={sp_feeder: inp})
for inp in input_scattered_cols_non_duplicate)
true_loss = self.calculate_loss_from_wals_model(wals_model,
self._wals_inputs)
self.assertNear(
loss,
true_loss,
err=.001,
msg="After col update, computed loss [{}] does not match"
" true loss [{}]".format(loss, true_loss))
def _run_test_process_input(self,
use_factors_weights_cache,
compute_loss=False):
with ops.Graph().as_default(), self.cached_session() as sess:
self._wals_inputs = self.sparse_input()
sp_feeder = array_ops.sparse_placeholder(dtypes.float32)
num_rows = 5
num_cols = 7
factor_dim = 3
wals_model = factorization_ops.WALSModel(
num_rows,
num_cols,
factor_dim,
num_row_shards=2,
num_col_shards=3,
regularization=0.01,
unobserved_weight=0.1,
col_init=self.col_init,
row_weights=self.row_wts,
col_weights=self.col_wts,
use_factors_weights_cache=use_factors_weights_cache)
wals_model.initialize_op.run()
wals_model.worker_init.run()
# Split input into multiple sparse tensors with scattered rows. Note that
# this split can be different than the factor sharding and the inputs can
# consist of non-consecutive rows. Each row needs to include all non-zero
# elements in that row.
sp_r0 = np_matrix_to_tf_sparse(INPUT_MATRIX, [0, 2]).eval()
sp_r1 = np_matrix_to_tf_sparse(INPUT_MATRIX, [1, 4], shuffle=True).eval()
sp_r2 = np_matrix_to_tf_sparse(INPUT_MATRIX, [3], shuffle=True).eval()
input_scattered_rows = [sp_r0, sp_r1, sp_r2]
# Test updating row factors.
# Here we feed in scattered rows of the input.
wals_model.row_update_prep_gramian_op.run()
wals_model.initialize_row_update_op.run()
(_, process_input_op, unregularized_loss, regularization,
_) = wals_model.update_row_factors(
sp_input=sp_feeder, transpose_input=False)
factor_loss = unregularized_loss + regularization
for inp in input_scattered_rows:
feed_dict = {sp_feeder: inp}
process_input_op.run(feed_dict=feed_dict)
row_factors = [x.eval() for x in wals_model.row_factors]
self.assertAllClose(row_factors[0], self._row_factors_0, atol=1e-3)
self.assertAllClose(row_factors[1], self._row_factors_1, atol=1e-3)
# Test row projection.
# Using the specified projection weights for the 2 row feature vectors.
# This is expected to reproduce the same row factors in the model as the
# weights and feature vectors are identical to that used in model
# training.
projected_rows = wals_model.project_row_factors(
sp_input=sp_feeder,
transpose_input=False,
projection_weights=[0.2, 0.5])
# Don't specify the projection weight, so 1.0 will be used. The feature
# weights will be those specified in model.
projected_rows_no_weights = wals_model.project_row_factors(
sp_input=sp_feeder, transpose_input=False)
feed_dict = {
sp_feeder:
np_matrix_to_tf_sparse(INPUT_MATRIX, [1, 4], shuffle=False)
.eval()
}
self.assertAllClose(
projected_rows.eval(feed_dict=feed_dict),
[self._row_factors_0[1], self._row_factors_1[1]],
atol=1e-3)
self.assertAllClose(
projected_rows_no_weights.eval(feed_dict=feed_dict),
[[0.569082, 0.715088, 0.31777], [1.915879, 1.992677, 1.109057]],
atol=1e-3)
if compute_loss:
# Test loss computation after the row update
loss = sum(
sess.run(
factor_loss * self.count_rows(inp) / num_rows,
feed_dict={sp_feeder: inp}) for inp in input_scattered_rows)
true_loss = self.calculate_loss_from_wals_model(wals_model,
self._wals_inputs)
self.assertNear(
loss,
true_loss,
err=.001,
msg="After row update, computed loss [{}] does not match"
" true loss [{}]".format(loss, true_loss))
# Split input into multiple sparse tensors with scattered columns. Note
# that here the elements in the sparse tensors are not ordered and also
# do not need to consist of consecutive columns. However, each column
# needs to include all non-zero elements in that column.
sp_c0 = np_matrix_to_tf_sparse(INPUT_MATRIX, col_slices=[2, 0]).eval()
sp_c1 = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[5, 3, 1], shuffle=True).eval()
sp_c2 = np_matrix_to_tf_sparse(INPUT_MATRIX, col_slices=[4, 6]).eval()
sp_c3 = np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[3, 6], shuffle=True).eval()
input_scattered_cols = [sp_c0, sp_c1, sp_c2, sp_c3]
input_scattered_cols_non_duplicate = [sp_c0, sp_c1, sp_c2]
# Test updating column factors.
# Here we feed in scattered columns of the input.
wals_model.col_update_prep_gramian_op.run()
wals_model.initialize_col_update_op.run()
(_, process_input_op, unregularized_loss, regularization,
_) = wals_model.update_col_factors(
sp_input=sp_feeder, transpose_input=False)
factor_loss = unregularized_loss + regularization
for inp in input_scattered_cols:
feed_dict = {sp_feeder: inp}
process_input_op.run(feed_dict=feed_dict)
col_factors = [x.eval() for x in wals_model.col_factors]
self.assertAllClose(col_factors[0], self._col_factors_0, atol=1e-3)
self.assertAllClose(col_factors[1], self._col_factors_1, atol=1e-3)
self.assertAllClose(col_factors[2], self._col_factors_2, atol=1e-3)
# Test column projection.
# Using the specified projection weights for the 3 column feature vectors.
# This is expected to reproduce the same column factors in the model as
# the weights and feature vectors are identical to that used in model
# training.
projected_cols = wals_model.project_col_factors(
sp_input=sp_feeder,
transpose_input=False,
projection_weights=[0.6, 0.4, 0.2])
# Don't specify the projection weight, so 1.0 will be used. The feature
# weights will be those specified in model.
projected_cols_no_weights = wals_model.project_col_factors(
sp_input=sp_feeder, transpose_input=False)
feed_dict = {
sp_feeder:
np_matrix_to_tf_sparse(
INPUT_MATRIX, col_slices=[5, 3, 1], shuffle=False).eval()
}
self.assertAllClose(
projected_cols.eval(feed_dict=feed_dict), [
self._col_factors_2[0], self._col_factors_1[0],
self._col_factors_0[1]
],
atol=1e-3)
self.assertAllClose(
projected_cols_no_weights.eval(feed_dict=feed_dict),
[[3.471045, -1.250835, -3.598917], [3.585139, -0.487476, -3.852232],
[0.346433, 1.360644, 1.677121]],
atol=1e-3)
if compute_loss:
# Test loss computation after the column update.
loss = sum(
sess.run(
factor_loss * self.count_cols(inp) / num_cols,
feed_dict={sp_feeder: inp})
for inp in input_scattered_cols_non_duplicate)
true_loss = self.calculate_loss_from_wals_model(wals_model,
self._wals_inputs)
self.assertNear(
loss,
true_loss,
err=.001,
msg="After col update, computed loss [{}] does not match"
" true loss [{}]".format(loss, true_loss))