def get_timing_signal_1d_given_position(channels,
position,
min_timescale=1.0,
max_timescale=1.0e4):
"""Get sinusoids of diff frequencies, with timing position given.
Adapted from add_timing_signal_1d_given_position in
//third_party/py/tensor2tensor/layers/common_attention.py
Args:
channels: scalar, size of timing embeddings to create. The number of
different timescales is equal to channels / 2.
position: a Tensor with shape [batch, seq_len]
min_timescale: a float
max_timescale: a float
Returns:
a Tensor of timing signals [batch, seq_len, channels]
"""
num_timescales = channels // 2
log_timescale_increment = (
math.log(float(max_timescale) / float(min_timescale)) /
(tf.to_float(num_timescales) - 1))
inv_timescales = min_timescale * tf.exp(
tf.to_float(tf.range(num_timescales)) * -log_timescale_increment)
scaled_time = (tf.expand_dims(tf.to_float(position), 2) *
tf.expand_dims(tf.expand_dims(inv_timescales, 0), 0))
signal = tf.concat([tf.sin(scaled_time), tf.cos(scaled_time)], axis=2)
signal = tf.pad(signal, [[0, 0], [0, 0], [0, tf.mod(channels, 2)]])
return signal
def get_tsa_threshold(tsa_schedule, global_step, num_train_steps, start, end):
training_progress = tf.to_float(global_step) / tf.to_float(num_train_steps)
if tsa_schedule == 'linear':
threshold = training_progress
elif tsa_schedule == 'exp':
scale = 5
threshold = tf.exp((training_progress - 1) * scale)
# [exp(-5), exp(0)] = [1e-2, 1]
elif tsa_schedule == 'log':
scale = 5
# [1 - exp(0), 1 - exp(-5)] = [0, 0.99]
threshold = 1 - tf.exp((-training_progress) * scale)
else:
raise ValueError(
'Invalid value for `tsa_schedule`: %s. Pick one from `linear`, '
'`exp` or `log`.' % (tsa_schedule))
return threshold * (end - start) + start
def mask(inputs, key_masks=None, type=None):
'''Masks paddings on keys or queries to inputs
inputs: 3d tensor. (h*N, T_q, T_k)
key_masks: 3d tensor. (N, 1, T_k)
type: string. 'key' | 'future'
e.g.,
>> inputs = tf.zeros([2, 2, 3], dtype=tf.float32)
>> key_masks = tf.constant([[0., 0., 1.],
[0., 1., 1.]])
>> mask(inputs, key_masks=key_masks, type='key')
array([[[ 0.0000000e+00, 0.0000000e+00, -4.2949673e+09],
[ 0.0000000e+00, 0.0000000e+00, -4.2949673e+09]],
[[ 0.0000000e+00, -4.2949673e+09, -4.2949673e+09],
[ 0.0000000e+00, -4.2949673e+09, -4.2949673e+09]],
[[ 0.0000000e+00, 0.0000000e+00, -4.2949673e+09],
[ 0.0000000e+00, 0.0000000e+00, -4.2949673e+09]],
[[ 0.0000000e+00, -4.2949673e+09, -4.2949673e+09],
[ 0.0000000e+00, -4.2949673e+09, -4.2949673e+09]]], dtype=float32)
'''
padding_num = -2 ** 32 + 1
if type in ('k', 'key', 'keys'):
key_masks = tf.to_float(key_masks)
key_masks = tf.tile(
key_masks,
[tf.shape(inputs)[0] // tf.shape(key_masks)[0], 1]) # (h*N, seqlen)
key_masks = tf.expand_dims(key_masks, 1) # (h*N, 1, seqlen)
outputs = inputs + key_masks * padding_num
# elif type in ('q', 'query', 'queries'):
# # Generate masks
# masks = tf.sign(tf.reduce_sum(tf.abs(queries), axis=-1)) # (N, T_q)
# masks = tf.expand_dims(masks, -1) # (N, T_q, 1)
# masks = tf.tile(masks, [1, 1, tf.shape(keys)[1]]) # (N, T_q, T_k)
#
# # Apply masks to inputs
# outputs = inputs*masks
elif type in ('f', 'future', 'right'):
diag_vals = tf.ones_like(inputs[0, :, :]) # (T_q, T_k)
tril = tf.linalg.LinearOperatorLowerTriangular(
diag_vals).to_dense() # (T_q, T_k)
future_masks = tf.tile(
tf.expand_dims(tril, 0),
[tf.shape(inputs)[0], 1, 1]) # (N, T_q, T_k)
paddings = tf.ones_like(future_masks) * padding_num
outputs = tf.where(tf.equal(future_masks, 0), paddings, inputs)
else:
print('Check if you entered type correctly!')
return outputs
def positional_encoding(inputs,
maxlen,
masking=True,
scope='positional_encoding'):
'''Sinusoidal Positional_Encoding. See 3.5
inputs: 3d tensor. (N, T, E)
maxlen: scalar. Must be >= T
masking: Boolean. If True, padding positions are set to zeros.
scope: Optional scope for `variable_scope`.
returns
3d tensor that has the same shape as inputs.
'''
E = inputs.get_shape().as_list()[-1] # static
N, T = tf.shape(inputs)[0], tf.shape(inputs)[1] # dynamic
with tf.variable_scope(scope, reuse=tf.AUTO_REUSE):
# position indices
position_ind = tf.tile(tf.expand_dims(tf.range(T), 0), [N, 1]) # (N, T)
# First part of the PE function: sin and cos argument
position_enc = np.array([
[pos / np.power(10000, (i-i%2)/E) for i in range(E)]
for pos in range(maxlen)])
# Second part, apply the cosine to even columns and sin to odds.
position_enc[:, 0::2] = np.sin(position_enc[:, 0::2]) # dim 2i
position_enc[:, 1::2] = np.cos(position_enc[:, 1::2]) # dim 2i+1
position_enc = tf.convert_to_tensor(
position_enc, tf.float32) # (maxlen, E)
# lookup
outputs = tf.nn.embedding_lookup(position_enc, position_ind)
# masks
if masking:
outputs = tf.where(tf.equal(inputs, 0), inputs, outputs)
return tf.to_float(outputs)