def test_transform_time_act_stride_unequal_batch_size():
# 2 strides batch size 3
acts = [[[[ 6.], [ 12.], [ 18.], [ 24.], [ 30.], [ 36.], [ 42.]]],
[[[ 51.], [ 57.], [ 63.], [ 69.], [ 75.], [ 81.], [ 87.]]],
[[[ -6.], [ -12.], [ -18.], [ -24.], [ -30.], [ -36.], [ -42.]]],
[[[ 9.], [ 15.], [ 21.], [ 27.], [ 33.], [ 39.], [ np.nan]]],
[[[ 54.], [ 60.], [ 66.], [ 72.], [ 78.], [ 84.], [ np.nan]]],
[[[ -9.], [ -15.], [ -21.], [ -27.], [ -33.], [ -39.], [ np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [3])
assert equal_without_nans(time_act, np.array(
[[[[ 6., 9., 12., 15., 18.,
21., 24., 27., 30., 33., 36.,
39., 42., np.nan]],
[[ 51., 54., 57., 60., 63., 66., 69., 72., 75., 78., 81.,
84., 87., np.nan]],
[[ -6., -9., -12., -15., -18., -21., -24., -27., -30., -33., -36.,
-39., -42., np.nan]]]]))
# 3 strides batch size 2
acts = [[[[ 6.], [ 12.], [ 18.], [ 24.], [ 30.], [ 36.], [ 42.]]],
[[[ 51.], [ 57.], [ 63.], [ 69.], [ 75.], [ 81.], [ 87.]]],
[[[ 9.], [ 15.], [ 21.], [ 27.], [ 33.], [ 39.],[45.]]],
[[[ 54.], [ 60.], [ 66.], [ 72.], [ 78.], [ 84.], [ 92.]]],
[[[ 10.], [ 16.], [ 22.], [ 28.], [ 34.], [ 40.], [ np.nan]]],
[[[ 55.], [ 61.], [ 67.], [ 73.], [ 79.], [ 85.], [ np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [2])
assert equal_without_nans(time_act, np.array([[[[ 6., 9., 10., 12., 15., 16., 18., 21., 22., 24., 27.,
28., 30., 33., 34., 36., 39., 40., 42., 45., np.nan]],
[[ 51., 54., 55., 57., 60., 61., 63., 66., 67., 69., 72.,
73., 75., 78., 79., 81., 84., 85., 87., 92., np.nan]],
]]))
def get_trial_acts(all_outs_per_batch, batch_sizes, n_trials, n_inputs_per_trial,
n_trial_len, n_sample_preds):
"""Compute trial activations from activations of a specific layer.
Parameters
----------
all_outs_per_batch: list of 1darray
All activations of a specific layer for all batches from the iterator.
batch_sizes: list
All batch sizes of all batches.
n_trials: int
n_inputs_per_trial: int
How many inputs/rows are used to predict all samples of one trial.
Depends on trial length, number of predictions per input window.
n_trial_len: int
Number of samples per trial
n_sample_preds: int
Number of predictions per input window.
Returns
--------
trial_acts: 3darray (final empty dim removed)
Activations of this layer for all trials.
"""
time_acts = transform_to_time_activations(all_outs_per_batch,batch_sizes)
trial_batch_acts = np.concatenate(time_acts, axis=0).reshape(n_trials,n_inputs_per_trial,
time_acts[0].shape[1], time_acts[0].shape[2], 1)
trial_acts = [transform_to_cnt_activations(t[np.newaxis], n_sample_preds,
n_samples = n_trial_len)
for t in trial_batch_acts]
trial_acts = np.array(trial_acts)
return trial_acts
def test_transform_time_cnt_act():
acts = [[[[ 6.], [ 12.], [ 18.], [ 24.], [ 30.], [ 36.], [ 42.]]],
[[[ 51.], [ 57.], [ 63.], [ 69.], [ 75.], [ 81.], [ 87.]]],
[[[ 9.], [ 15.], [ 21.], [ 27.], [ 33.], [ 39.], [ np.nan]]],
[[[ 54.], [ 60.], [ 66.], [ 72.], [ 78.], [ 84.], [ np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [2])
assert len(time_act) == 1
assert equal_without_nans(np.array([[[ 6., 9., 12., 15., 18.,
21., 24., 27., 30., 33., 36.,
39., 42., np.nan]],
[[ 51., 54., 57., 60., 63., 66., 69., 72., 75., 78., 81.,
84., 87., np.nan]]]),time_act[0])
cnt_act = transform_to_cnt_activations(time_act, n_sample_preds=7, n_samples=14)
assert np.array_equal([[ 24., 27., 30., 33., 36., 39., 42., 69., 72., 75., 78.,
81., 84., 87.]], cnt_act)
cnt_act = transform_to_cnt_activations(time_act, n_sample_preds=9, n_samples=18)
assert np.array_equal([[ 18.,21, 24., 27., 30., 33., 36., 39., 42.,
63, 66, 69., 72., 75., 78.,
81., 84., 87.]], cnt_act)
cnt_act = transform_to_cnt_activations(time_act, n_sample_preds=13, n_samples=26)
assert np.array_equal([[ 6, 9, 12, 15, 18.,21, 24., 27., 30.,
33., 36., 39., 42.,
51, 54 ,57, 60, 63, 66, 69., 72., 75., 78.,
81., 84., 87.]], cnt_act)
def test_transform_time_act_stride_unequal_batch_size():
# 2 strides batch size 3
acts = [[[[6.], [12.], [18.], [24.], [30.], [36.], [42.]]],
[[[51.], [57.], [63.], [69.], [75.], [81.], [87.]]],
[[[-6.], [-12.], [-18.], [-24.], [-30.], [-36.], [-42.]]],
[[[9.], [15.], [21.], [27.], [33.], [39.], [np.nan]]],
[[[54.], [60.], [66.], [72.], [78.], [84.], [np.nan]]],
[[[-9.], [-15.], [-21.], [-27.], [-33.], [-39.], [np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [3])
assert equal_without_nans(
time_act,
np.array([[[[
6., 9., 12., 15., 18., 21., 24., 27., 30., 33., 36., 39., 42.,
np.nan
]],
[[
51., 54., 57., 60., 63., 66., 69., 72., 75., 78., 81.,
84., 87., np.nan
]],
[[
-6., -9., -12., -15., -18., -21., -24., -27., -30.,
-33., -36., -39., -42., np.nan
]]]]))
# 3 strides batch size 2
acts = [[[[6.], [12.], [18.], [24.], [30.], [36.], [42.]]],
[[[51.], [57.], [63.], [69.], [75.], [81.], [87.]]],
[[[9.], [15.], [21.], [27.], [33.], [39.], [45.]]],
[[[54.], [60.], [66.], [72.], [78.], [84.], [92.]]],
[[[10.], [16.], [22.], [28.], [34.], [40.], [np.nan]]],
[[[55.], [61.], [67.], [73.], [79.], [85.], [np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [2])
assert equal_without_nans(
time_act,
np.array([[
[[
6., 9., 10., 12., 15., 16., 18., 21., 22., 24., 27., 28., 30.,
33., 34., 36., 39., 40., 42., 45., np.nan
]],
[[
51., 54., 55., 57., 60., 61., 63., 66., 67., 69., 72., 73.,
75., 78., 79., 81., 84., 85., 87., 92., np.nan
]],
]]))
def test_transform_time_cnt_act_multiple_chans():
# two batches, first with two trials,
# second with one trial
before = np.array([
[[[1, 4, 7], [-1, -4, -7]], [[11, 14, 17], [-11, -14, -17]],
[[2, 5, 8], [-2, -5, -8]], [[12, 15, 18], [-12, -15, -18]],
[[3, 6, 9], [-3, -6, -9]], [[13, 16, 19], [-13, -16, -19]]],
# second batch
[[[21, 24, 27], [-21, -24, -27]], [[22, 25, 28], [-22, -25, -28]],
[[23, 26, 29], [-23, -26, -29]]]
])
before[0] = np.array(before[0])[:, :, :, np.newaxis]
before[1] = np.array(before[1])[:, :, :, np.newaxis]
time_act = transform_to_time_activations(before, [2, 1])
assert np.array_equal(
np.array([[[1, 2, 3, 4, 5, 6, 7, 8, 9],
[-1, -2, -3, -4, -5, -6, -7, -8, -9]],
[[11, 12, 13, 14, 15, 16, 17, 18, 19],
[-11, -12, -13, -14, -15, -16, -17, -18, -19]]]),
time_act[0])
assert np.array_equal(
np.array([[[21, 22, 23, 24, 25, 26, 27, 28, 29],
[-21, -22, -23, -24, -25, -26, -27, -28, -29]]]),
time_act[1])
cnt_act = transform_to_cnt_activations(time_act,
n_sample_preds=5,
n_samples=15)
assert np.array_equal(
np.array([[5, 6, 7, 8, 9, 15, 16, 17, 18, 19, 25, 26, 27, 28, 29],
[
-5, -6, -7, -8, -9, -15, -16, -17, -18, -19, -25, -26,
-27, -28, -29
]]), cnt_act)
cnt_act = transform_to_cnt_activations(time_act,
n_sample_preds=9,
n_samples=27)
assert np.array_equal(
np.array([[
1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16, 17, 18, 19, 21,
22, 23, 24, 25, 26, 27, 28, 29
],
[
-1, -2, -3, -4, -5, -6, -7, -8, -9, -11, -12, -13, -14,
-15, -16, -17, -18, -19, -21, -22, -23, -24, -25, -26,
-27, -28, -29
]]), cnt_act)
def test_transform_time_cnt_act_multiple_chans():
# two batches, first with two trials,
# second with one trial
before = np.array([
[[[1, 4, 7], [-1, -4, -7]], [[11, 14, 17], [-11, -14, -17]],
[[2, 5, 8], [-2, -5, -8]], [[12, 15, 18], [-12, -15, -18]],
[[3, 6, 9], [-3, -6, -9]], [[13, 16, 19], [-13, -16, -19]]],
# second batch
[[[21, 24, 27], [-21, -24, -27]], [[22, 25, 28], [-22, -25, -28]],
[[23, 26, 29], [-23, -26, -29]]]
])
before[0] = np.array(before[0])[:, :, :, np.newaxis]
before[1] = np.array(before[1])[:, :, :, np.newaxis]
time_act = transform_to_time_activations(before, [2, 1])
assert np.array_equal(
np.array([[[1, 2, 3, 4, 5, 6, 7, 8, 9],
[-1, -2, -3, -4, -5, -6, -7, -8, -9]],
[[11, 12, 13, 14, 15, 16, 17, 18, 19],
[-11, -12, -13, -14, -15, -16, -17, -18, -19]]]),
time_act[0])
assert np.array_equal(
np.array([[[21, 22, 23, 24, 25, 26, 27, 28, 29],
[-21, -22, -23, -24, -25, -26, -27, -28, -29]]]),
time_act[1])
cnt_act = transform_to_cnt_activations(
time_act, n_sample_preds=5, n_samples=15)
assert np.array_equal(
np.array([[5, 6, 7, 8, 9, 15, 16, 17, 18, 19, 25, 26, 27, 28, 29],
[
-5, -6, -7, -8, -9, -15, -16, -17, -18, -19, -25, -26,
-27, -28, -29
]]), cnt_act)
cnt_act = transform_to_cnt_activations(
time_act, n_sample_preds=9, n_samples=27)
assert np.array_equal(
np.array([[
1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16, 17, 18, 19, 21,
22, 23, 24, 25, 26, 27, 28, 29
],
[
-1, -2, -3, -4, -5, -6, -7, -8, -9, -11, -12, -13, -14,
-15, -16, -17, -18, -19, -21, -22, -23, -24, -25, -26,
-27, -28, -29
]]), cnt_act)
def test_transform_time_cnt_act():
acts = [[[[6.], [12.], [18.], [24.], [30.], [36.], [42.]]],
[[[51.], [57.], [63.], [69.], [75.], [81.], [87.]]],
[[[9.], [15.], [21.], [27.], [33.], [39.], [np.nan]]],
[[[54.], [60.], [66.], [72.], [78.], [84.], [np.nan]]]]
time_act = transform_to_time_activations(np.array([acts]), [2])
assert len(time_act) == 1
assert equal_without_nans(
np.array([[[
6., 9., 12., 15., 18., 21., 24., 27., 30., 33., 36., 39., 42.,
np.nan
]],
[[
51., 54., 57., 60., 63., 66., 69., 72., 75., 78., 81.,
84., 87., np.nan
]]]), time_act[0])
cnt_act = transform_to_cnt_activations(time_act,
n_sample_preds=7,
n_samples=14)
assert np.array_equal([[
24., 27., 30., 33., 36., 39., 42., 69., 72., 75., 78., 81., 84., 87.
]], cnt_act)
cnt_act = transform_to_cnt_activations(time_act,
n_sample_preds=9,
n_samples=18)
assert np.array_equal([[
18., 21, 24., 27., 30., 33., 36., 39., 42., 63, 66, 69., 72., 75., 78.,
81., 84., 87.
]], cnt_act)
cnt_act = transform_to_cnt_activations(time_act,
n_sample_preds=13,
n_samples=26)
assert np.array_equal([[
6, 9, 12, 15, 18., 21, 24., 27., 30., 33., 36., 39., 42., 51, 54, 57,
60, 63, 66, 69., 72., 75., 78., 81., 84., 87.
]], cnt_act)
def get_trial_acts(all_outs_per_batch, batch_sizes, n_trials,
n_inputs_per_trial, n_trial_len, n_sample_preds):
"""Compute trial activations from activations of a specific layer.
Parameters
----------
all_outs_per_batch: list of 1darray
All activations of a specific layer for all batches from the iterator.
batch_sizes: list
All batch sizes of all batches.
n_trials: int
n_inputs_per_trial: int
How many inputs/rows are used to predict all samples of one trial.
Depends on trial length, number of predictions per input window.
n_trial_len: int
Number of samples per trial
n_sample_preds: int
Number of predictions per input window.
Returns
--------
trial_acts: 3darray (final empty dim removed)
Activations of this layer for all trials.
"""
time_acts = transform_to_time_activations(all_outs_per_batch, batch_sizes)
trial_batch_acts = np.concatenate(time_acts,
axis=0).reshape(n_trials,
n_inputs_per_trial,
time_acts[0].shape[1],
time_acts[0].shape[2], 1)
trial_acts = [
transform_to_cnt_activations(t[np.newaxis],
n_sample_preds,
n_samples=n_trial_len)
for t in trial_batch_acts
]
trial_acts = np.array(trial_acts)
return trial_acts