def test_dynamic_api(self):
paddle.disable_static()
x = paddle.to_tensor(self.input_x)
y = paddle.to_tensor(self.input_y)
z = paddle.to_tensor(self.input_z)
a = paddle.to_tensor(self.input_a)
b = paddle.to_tensor(self.input_b)
c = paddle.to_tensor(self.input_c)
res = paddle.subtract(x, y)
res = res.numpy()
self.assertTrue(np.allclose(res, self.np_expected1))
# test broadcast
res = paddle.subtract(x, z)
res = res.numpy()
self.assertTrue(np.allclose(res, self.np_expected2))
res = paddle.subtract(a, c)
res = res.numpy()
self.assertTrue(np.allclose(res, self.np_expected3))
res = paddle.subtract(b, c)
res = res.numpy()
self.assertTrue(np.allclose(res, self.np_expected4))
def build_program():
main_program = paddle.static.Program()
startup_program = paddle.static.Program()
with paddle.static.program_guard(main_program, startup_program):
with paddle.static.device_guard('cpu'):
data = paddle.ones([4, 64], dtype='float32', name='data')
# data -> [memcpy_h2d] -> data' -> [matmul] -> out ->[add] -> add_out
with paddle.static.device_guard('gpu'):
weight = paddle.randn([64, 64], name='weight') # gpu
matmul_out = paddle.matmul(data, weight, name='matmul_out') # gpus
bias = paddle.ones([4, 64], dtype='float32', name='bias')
add_out = paddle.add(matmul_out, bias, name='add_out')
# add_out -> [memcpy_d2h] -> add_out' -> [sub] -> sub_out -> [tanh] -> tanh_out
with paddle.static.device_guard('cpu'):
sub_out = paddle.subtract(add_out, data, name='sub_out')
tanh_out = paddle.tanh(sub_out, name='tanh_out')
with paddle.static.device_guard('gpu'):
bias_1 = paddle.add(bias, sub_out, name='bias_1')
out_before = paddle.tanh(bias_1, name='out_before')
out_last = paddle.subtract(tanh_out, data, name='out_last')
out = paddle.add(out_before, out_last, name='out')
mean = paddle.mean(out, name='mean_out')
return main_program, startup_program, [mean]
def test_static_api(self):
paddle.enable_static()
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
data_x = paddle.static.data("x", shape=[10, 15], dtype="float32")
data_y = paddle.static.data("y", shape=[10, 15], dtype="float32")
result_max = paddle.subtract(data_x, data_y)
exe = paddle.static.Executor(self.place)
res, = exe.run(feed={
"x": self.input_x,
"y": self.input_y
},
fetch_list=[result_max])
self.assertTrue(np.allclose(res, self.np_expected1))
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
data_x = paddle.static.data("x", shape=[10, 15], dtype="float32")
data_z = paddle.static.data("z", shape=[15], dtype="float32")
result_max = paddle.subtract(data_x, data_z)
exe = paddle.static.Executor(self.place)
res, = exe.run(feed={
"x": self.input_x,
"z": self.input_z
},
fetch_list=[result_max])
self.assertTrue(np.allclose(res, self.np_expected2))
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
data_a = paddle.static.data("a", shape=[3], dtype="int64")
data_c = paddle.static.data("c", shape=[3], dtype="int64")
result_max = paddle.subtract(data_a, data_c)
exe = paddle.static.Executor(self.place)
res, = exe.run(feed={
"a": self.input_a,
"c": self.input_c
},
fetch_list=[result_max])
self.assertTrue(np.allclose(res, self.np_expected3))
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
data_b = paddle.static.data("b", shape=[3], dtype="int64")
data_c = paddle.static.data("c", shape=[3], dtype="int64")
result_max = paddle.subtract(data_b, data_c)
exe = paddle.static.Executor(self.place)
res, = exe.run(feed={
"b": self.input_b,
"c": self.input_c
},
fetch_list=[result_max])
self.assertTrue(np.allclose(res, self.np_expected4))
def test_name(self):
with paddle.static.program_guard(paddle.static.Program()):
x = paddle.static.data(name="x", shape=[2, 3], dtype="float32")
y = paddle.static.data(name='y', shape=[2, 3], dtype='float32')
y_1 = paddle.subtract(x, y, name='add_res')
self.assertEqual(('add_res' in y_1.name), True)
def net(self, inputs, is_infer=False):
pyramid_model = MatchPyramidLayer(
self.emb_path, self.vocab_size, self.emb_size, self.kernel_num,
self.conv_filter, self.conv_act, self.hidden_size, self.out_size,
self.pool_size, self.pool_stride, self.pool_padding,
self.pool_type, self.hidden_act)
prediction = pyramid_model(inputs)
if is_infer:
self._infer_results["prediction"] = prediction
return
pos = paddle.slice(
prediction, axes=[0, 1], starts=[0, 0], ends=[64, 1])
neg = paddle.slice(
prediction, axes=[0, 1], starts=[64, 0], ends=[128, 1])
loss_part1 = paddle.subtract(
paddle.full(
shape=[64, 1], fill_value=1.0, dtype='float32'), pos)
loss_part2 = paddle.add(loss_part1, neg)
loss_part3 = paddle.maximum(
paddle.full(
shape=[64, 1], fill_value=0.0, dtype='float32'),
loss_part2)
avg_cost = paddle.mean(loss_part3)
self._cost = avg_cost
def test_static(self):
with paddle.static.program_guard(paddle.static.Program()):
x_np = np.array([2, 3, 4]).astype('float32')
y_np = np.array([1, 5, 2]).astype('float32')
x = paddle.static.data(name="x", shape=[3], dtype='float32')
y = paddle.static.data(name="y", shape=[3], dtype='float32')
x_reshape = paddle.reshape(x, [3, 1])
y_reshape = paddle.reshape(y, [3, 1])
z = paddle.subtract(x_reshape, y_reshape)
z = paddle.reshape(z, shape=[3])
place = paddle.NPUPlace(0)
exe = paddle.static.Executor(place)
x_value, y_value, z_value = exe.run(feed={
"x": x_np,
"y": y_np
},
fetch_list=[x, y, z])
z_expected = np.array([1., -2., 2.])
self.assertEqual(
(x_value == x_np).all(),
True,
msg="x_value = {}, but expected {}".format(x_value, x_np))
self.assertEqual(
(y_value == y_np).all(),
True,
msg="y_value = {}, but expected {}".format(y_value, y_np))
self.assertEqual((z_value == z_expected).all(),
True,
msg="z_value = {}, but expected {}".format(
z_value, z_expected))
def forward(self, x, y):
if in_dygraph_mode():
sub = _C_ops.elementwise_sub(x, y)
return _C_ops.final_state_p_norm(sub, self.p, 1, self.epsilon,
self.keepdim, False)
if _in_legacy_dygraph():
sub = _C_ops.elementwise_sub(x, y)
return _C_ops.p_norm(sub, 'axis', 1, 'porder', self.p, 'keepdim',
self.keepdim, 'epsilon', self.epsilon)
check_variable_and_dtype(x, 'x', ['float32', 'float64'],
'PairwiseDistance')
check_variable_and_dtype(y, 'y', ['float32', 'float64'],
'PairwiseDistance')
sub = paddle.subtract(x, y)
helper = LayerHelper("PairwiseDistance", name=self.name)
attrs = {
'axis': 1,
'porder': self.p,
'keepdim': self.keepdim,
'epsilon': self.epsilon,
}
out = helper.create_variable_for_type_inference(dtype=x.dtype)
helper.append_op(type='p_norm',
inputs={'X': sub},
outputs={'Out': out},
attrs=attrs)
return out
def net(self, input, is_infer=False):
pyramid_model = MatchPyramidLayer(
self.emb_path, self.vocab_size, self.emb_size, self.kernel_num,
self.conv_filter, self.conv_act, self.hidden_size, self.out_size,
self.pool_size, self.pool_stride, self.pool_padding,
self.pool_type, self.hidden_act)
prediction = pyramid_model(input)
if is_infer:
fetch_dict = {'prediction': prediction}
return fetch_dict
# calculate hinge loss
pos = paddle.slice(prediction,
axes=[0, 1],
starts=[0, 0],
ends=[64, 1])
neg = paddle.slice(prediction,
axes=[0, 1],
starts=[64, 0],
ends=[128, 1])
loss_part1 = paddle.subtract(
paddle.full(shape=[64, 1], fill_value=1.0, dtype='float32'), pos)
loss_part2 = paddle.add(loss_part1, neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[64, 1], fill_value=0.0, dtype='float32'),
loss_part2)
avg_cost = paddle.mean(loss_part3)
self.inference_target_var = avg_cost
self._cost = avg_cost
fetch_dict = {'cost': avg_cost}
return fetch_dict
def _layer_sub(inputs, node):
"""
layer_sub, input(-1, emb_size), node(-1, n, emb_size)
"""
input_re = paddle.unsqueeze(inputs, axis=[1])
sub_res = paddle.subtract(input_re, node)
return sub_res
def softmax_with_cross_entropy(self, shard_logit, shard_one_hot):
shard_max = paddle.max(shard_logit, axis=1, keepdim=True)
global_max = shard_max
paddle.distributed.all_reduce(global_max,
op=paddle.distributed.ReduceOp.MAX)
shard_logit_new = paddle.subtract(shard_logit, global_max)
shard_exp = paddle.exp(shard_logit_new)
shard_demon = paddle.sum(shard_exp, axis=1, keepdim=True)
global_demon = shard_demon
paddle.distributed.all_reduce(global_demon,
op=paddle.distributed.ReduceOp.SUM)
global_log_demon = paddle.log(global_demon)
shard_log_prob = shard_logit_new - global_log_demon
shard_prob = paddle.exp(shard_log_prob)
target_log_prob = paddle.min(shard_log_prob * shard_one_hot,
axis=1,
keepdim=True)
shard_loss = paddle.scale(target_log_prob, scale=-1.0)
#TODO paddle.distributed.reducescatter not found
global_loss = paddle.fluid.layers.collective._c_reducescatter(
shard_loss, nranks=self.nranks, use_calc_stream=True)
return global_loss, shard_prob
def forward(self):
input = self.input('Input', 0)
im_info = self.input('ImInfo', 0)
im_info = paddle.reshape(im_info, shape=[3])
h, w, s = paddle.tensor.split(im_info, axis=0, num_or_sections=3)
tensor_one = paddle.full(shape=[1], dtype='float32', fill_value=1.0)
tensor_zero = paddle.full(shape=[1], dtype='float32', fill_value=0.0)
h = paddle.subtract(h, tensor_one)
w = paddle.subtract(w, tensor_one)
xmin, ymin, xmax, ymax = paddle.tensor.split(input,
axis=-1,
num_or_sections=4)
xmin = paddle.maximum(paddle.minimum(xmin, w), tensor_zero)
ymin = paddle.maximum(paddle.minimum(ymin, h), tensor_zero)
xmax = paddle.maximum(paddle.minimum(xmax, w), tensor_zero)
ymax = paddle.maximum(paddle.minimum(ymax, h), tensor_zero)
cliped_box = paddle.concat([xmin, ymin, xmax, ymax], axis=-1)
return {'Output': [cliped_box]}
def create_loss(batch_size, margin, cos_pos, cos_neg):
loss_part1 = paddle.subtract(
paddle.full(shape=[batch_size, 1], fill_value=margin, dtype='float32'),
cos_pos)
loss_part2 = paddle.add(loss_part1, cos_neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[batch_size, 1], fill_value=0.0, dtype='float32'),
loss_part2)
avg_cost = paddle.mean(loss_part3)
return avg_cost
def create_loss(prediction):
pos = paddle.slice(prediction, axes=[0, 1], starts=[0, 0], ends=[64, 1])
neg = paddle.slice(prediction, axes=[0, 1], starts=[64, 0], ends=[128, 1])
loss_part1 = paddle.subtract(
paddle.full(shape=[64, 1], fill_value=1.0, dtype='float32'), pos)
loss_part2 = paddle.add(loss_part1, neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[64, 1], fill_value=0.0, dtype='float32'),
loss_part2)
avg_cost = paddle.mean(loss_part3)
return avg_cost
def _test(self, run_npu=True):
main_prog = paddle.static.Program()
startup_prog = paddle.static.Program()
main_prog.random_seed = SEED
startup_prog.random_seed = SEED
np.random.seed(SEED)
a_np = np.random.random(size=(32, 32)).astype('float32')
b_np = np.random.random(size=(32, 32)).astype('float32')
label_np = np.random.randint(2, size=(32, 1)).astype('int64')
with paddle.static.program_guard(main_prog, startup_prog):
a = paddle.static.data(name="a", shape=[32, 32], dtype='float32')
b = paddle.static.data(name="b", shape=[32, 32], dtype='float32')
label = paddle.static.data(name="label",
shape=[32, 1],
dtype='int64')
sum = paddle.add(a, b)
c = paddle.assign(b)
z = paddle.subtract(sum, c)
fc_1 = fluid.layers.fc(input=z, size=128)
prediction = fluid.layers.fc(input=fc_1, size=2, act='softmax')
cost = fluid.layers.cross_entropy(input=prediction, label=label)
loss = fluid.layers.reduce_mean(cost)
sgd = fluid.optimizer.SGD(learning_rate=0.01)
sgd.minimize(loss)
if run_npu:
place = paddle.NPUPlace(0)
else:
place = paddle.CPUPlace()
exe = paddle.static.Executor(place)
exe.run(startup_prog)
for epoch in range(100):
pred_res, loss_res = exe.run(main_prog,
feed={
"a": a_np,
"b": b_np,
"label": label_np
},
fetch_list=[prediction, loss])
if epoch % 10 == 0:
print("Epoch {} | Prediction[0]: {}, Loss: {}".format(
epoch, pred_res[0], loss_res))
return pred_res, loss_res
def _margin_softmax(input, label, out_dim, param_attr, margin1, margin2,
margin3, scale, sample_ratio):
input_norm = paddle.sqrt(
paddle.sum(paddle.square(input), axis=1, keepdim=True))
input = paddle.divide(input, input_norm)
if param_attr is None:
param_attr = paddle.ParamAttr(
initializer=paddle.nn.initializer.XavierNormal(fan_in=0.0))
weight = paddle.static.create_parameter(
shape=[input.shape[1], out_dim],
dtype='float32',
name=unique_name.generate('final_fc_w'),
attr=param_attr)
if sample_ratio < 1.0:
# partial fc sample process
label, sampled_class_index = class_center_sample(
label, out_dim, ratio=sample_ratio, ignore_label=-1)
sampled_class_index.stop_gradient = True
weight = paddle.gather(weight, sampled_class_index, axis=1)
out_dim = paddle.shape(sampled_class_index)
weight_norm = paddle.sqrt(
paddle.sum(paddle.square(weight), axis=0, keepdim=True))
weight = paddle.divide(weight, weight_norm)
cos = paddle.matmul(input, weight)
theta = paddle.acos(cos)
if margin1 != 1.0:
theta = margin1 * theta
if margin2 != 0.0:
theta = theta + margin2
margin_cos = paddle.cos(theta)
if margin3 != 0.0:
margin_cos = margin_cos - margin3
one_hot = paddle.nn.functional.one_hot(label, num_classes=out_dim)
diff = paddle.multiply(paddle.subtract(margin_cos, cos), one_hot)
target_cos = paddle.add(cos, diff)
logit = paddle.scale(target_cos, scale=scale)
loss, prob = paddle.nn.functional.softmax_with_cross_entropy(
logits=logit,
label=paddle.reshape(label, (-1, 1)),
return_softmax=True)
avg_loss = paddle.mean(x=loss)
one_hot.stop_gradient = True
return avg_loss, prob
def forward(self, predicts, batch):
if isinstance(predicts, (list, tuple)):
predicts = predicts[-1]
predicts = predicts.transpose((1, 0, 2))
N, B, _ = predicts.shape
preds_lengths = paddle.to_tensor([N] * B, dtype='int64')
labels = batch[1].astype("int32")
label_lengths = batch[2].astype('int64')
loss = self.loss_func(predicts, labels, preds_lengths, label_lengths)
if self.use_focal_loss:
weight = paddle.exp(-loss)
weight = paddle.subtract(paddle.to_tensor([1.0]), weight)
weight = paddle.square(weight)
loss = paddle.multiply(loss, weight)
loss = loss.mean()
return {'loss': loss}
def forward(self,
query_input_ids,
title_input_ids,
query_token_type_ids=None,
query_position_ids=None,
query_attention_mask=None,
title_token_type_ids=None,
title_position_ids=None,
title_attention_mask=None):
query_token_embedding, _ = self.ptm(query_input_ids,
query_token_type_ids,
query_position_ids,
query_attention_mask)
query_token_embedding = self.dropout(query_token_embedding)
query_attention_mask = paddle.unsqueeze(
(query_input_ids != self.ptm.pad_token_id).astype(
self.ptm.pooler.dense.weight.dtype),
axis=2)
# Set token embeddings to 0 for padding tokens
query_token_embedding = query_token_embedding * query_attention_mask
query_sum_embedding = paddle.sum(query_token_embedding, axis=1)
query_sum_mask = paddle.sum(query_attention_mask, axis=1)
query_mean = query_sum_embedding / query_sum_mask
title_token_embedding, _ = self.ptm(title_input_ids,
title_token_type_ids,
title_position_ids,
title_attention_mask)
title_token_embedding = self.dropout(title_token_embedding)
title_attention_mask = paddle.unsqueeze(
(title_input_ids != self.ptm.pad_token_id).astype(
self.ptm.pooler.dense.weight.dtype),
axis=2)
# Set token embeddings to 0 for padding tokens
title_token_embedding = title_token_embedding * title_attention_mask
title_sum_embedding = paddle.sum(title_token_embedding, axis=1)
title_sum_mask = paddle.sum(title_attention_mask, axis=1)
title_mean = title_sum_embedding / title_sum_mask
sub = paddle.abs(paddle.subtract(query_mean, title_mean))
projection = paddle.concat([query_mean, title_mean, sub], axis=-1)
logits = self.classifier(projection)
probs = F.softmax(logits)
return probs
def __call__(self, predicts, batch):
if isinstance(predicts, (list, tuple)):
predicts = predicts[-1]
B, N = predicts.shape[:2]
div = paddle.to_tensor([N]).astype('float32')
predicts = nn.functional.softmax(predicts, axis=-1)
aggregation_preds = paddle.sum(predicts, axis=1)
aggregation_preds = paddle.divide(aggregation_preds, div)
length = batch[2].astype("float32")
batch = batch[3].astype("float32")
batch[:, 0] = paddle.subtract(div, length)
batch = paddle.divide(batch, div)
loss = self.loss_func(aggregation_preds, batch)
return {"loss_ace": loss}
def net(self, input, is_infer=False):
""" network"""
if is_infer:
self.batch_size = envs.get_global_env(
"dataset.inferdata.batch_size")
else:
self.batch_size = envs.get_global_env(
"dataset.sample_1.batch_size")
tagspace_model = TagspaceLayer(self.vocab_text_size,
self.vocab_tag_size, self.emb_dim,
self.hid_dim, self.win_size,
self.margin, self.neg_size,
self.text_len)
cos_pos, cos_neg = tagspace_model(input)
# calculate hinge loss
loss_part1 = paddle.subtract(
paddle.full(shape=[self.batch_size, 1],
fill_value=self.margin,
dtype='float32'), cos_pos)
loss_part2 = paddle.add(loss_part1, cos_neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[self.batch_size, 1],
fill_value=0.0,
dtype='float32'), loss_part2)
avg_cost = paddle.mean(loss_part3)
less = paddle.cast(paddle.less_than(cos_neg, cos_pos), dtype='float32')
label_ones = paddle.full(dtype='float32',
shape=[self.batch_size, 1],
fill_value=1.0)
correct = paddle.sum(less)
total = paddle.sum(label_ones)
acc = paddle.divide(correct, total)
self._cost = avg_cost
if is_infer:
self._infer_results["acc"] = acc
self._infer_results["loss"] = self._cost
else:
self._metrics["acc"] = acc
self._metrics["loss"] = self._cost
def net(self, input, is_infer=False):
if is_infer:
self.batch_size = self.config.get("runner.infer_batch_size")
else:
self.batch_size = self.config.get("runner.train_batch_size")
tagspace_model = TagspaceLayer(self.vocab_text_size,
self.vocab_tag_size, self.emb_dim,
self.hid_dim, self.win_size,
self.margin, self.neg_size,
self.text_len)
cos_pos, cos_neg = tagspace_model(input)
# calculate hinge loss
loss_part1 = paddle.subtract(
paddle.full(shape=[self.batch_size, 1],
fill_value=self.margin,
dtype='float32'), cos_pos)
loss_part2 = paddle.add(loss_part1, cos_neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[self.batch_size, 1],
fill_value=0.0,
dtype='float32'), loss_part2)
avg_cost = paddle.mean(loss_part3)
less = paddle.cast(paddle.less_than(cos_neg, cos_pos), dtype='float32')
label_ones = paddle.full(dtype='float32',
shape=[self.batch_size, 1],
fill_value=1.0)
correct = paddle.sum(less)
total = paddle.sum(label_ones)
acc = paddle.divide(correct, total)
self.inference_target_var = acc
if is_infer:
fetch_dict = {'ACC': acc}
return fetch_dict
self._cost = avg_cost
fetch_dict = {'cost': avg_cost, 'ACC': acc}
return fetch_dict
def net(self, input, is_infer=False):
self.q_slots = self._sparse_data_var[0:1]
self.pt_slots = self._sparse_data_var[1:2]
if not is_infer:
self.batch_size = envs.get_global_env(
"dataset.dataset_train.batch_size")
self.nt_slots = self._sparse_data_var[2:3]
inputs = [self.q_slots, self.pt_slots, self.nt_slots]
else:
self.batch_size = envs.get_global_env(
"dataset.dataset_infer.batch_size")
inputs = [self.q_slots, self.pt_slots]
simnet_model = MultiviewSimnetLayer(
self.query_encoder, self.title_encoder, self.query_encode_dim,
self.title_encode_dim, self.emb_size, self.emb_dim,
self.hidden_size, self.margin, self.query_len, self.pos_len,
self.neg_len)
cos_pos, cos_neg = simnet_model(inputs, is_infer)
if is_infer:
self._infer_results['query_pt_sim'] = cos_pos
return
# pairwise hinge_loss
loss_part1 = paddle.subtract(
paddle.full(shape=[self.batch_size, 1],
fill_value=self.margin,
dtype='float32'), cos_pos)
loss_part2 = paddle.add(loss_part1, cos_neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[self.batch_size, 1],
fill_value=0.0,
dtype='float32'), loss_part2)
self._cost = paddle.mean(loss_part3)
self.acc = self.get_acc(cos_neg, cos_pos)
self._metrics["loss"] = self._cost
self._metrics["acc"] = self.acc
def forward(self, similarities_matrix, query_img_id, gallery_img_id,
keep_mask):
metric_dict = dict()
choosen_indices = paddle.argsort(similarities_matrix,
axis=1,
descending=True)
gallery_labels_transpose = paddle.transpose(gallery_img_id, [1, 0])
gallery_labels_transpose = paddle.broadcast_to(
gallery_labels_transpose,
shape=[
choosen_indices.shape[0], gallery_labels_transpose.shape[1]
])
choosen_label = paddle.index_sample(gallery_labels_transpose,
choosen_indices)
equal_flag = paddle.equal(choosen_label, query_img_id)
if keep_mask is not None:
keep_mask = paddle.index_sample(keep_mask.astype('float32'),
choosen_indices)
equal_flag = paddle.logical_and(equal_flag,
keep_mask.astype('bool'))
equal_flag = paddle.cast(equal_flag, 'float32')
num_rel = paddle.sum(equal_flag, axis=1)
num_rel = paddle.greater_than(num_rel, paddle.to_tensor(0.))
num_rel_index = paddle.nonzero(num_rel.astype("int"))
num_rel_index = paddle.reshape(num_rel_index, [num_rel_index.shape[0]])
equal_flag = paddle.index_select(equal_flag, num_rel_index, axis=0)
#do accumulative sum
div = paddle.arange(equal_flag.shape[1]).astype("float32") + 2
minus = paddle.divide(equal_flag, div)
auxilary = paddle.subtract(equal_flag, minus)
hard_index = paddle.argmax(auxilary, axis=1).astype("float32")
all_INP = paddle.divide(paddle.sum(equal_flag, axis=1), hard_index)
mINP = paddle.mean(all_INP)
metric_dict["mINP"] = mINP.numpy()[0]
return metric_dict
def net(self, input, is_infer=False):
self.q_slots = [input[0]]
self.pt_slots = [input[1]]
if not is_infer:
self.batch_size = self.config.get("runner.train_batch_size")
self.nt_slots = [input[2]]
inputs = [self.q_slots, self.pt_slots, self.nt_slots]
else:
self.batch_size = self.config.get("runner.infer_batch_size")
inputs = [self.q_slots, self.pt_slots]
simnet_model = MultiviewSimnetLayer(
self.query_encoder, self.title_encoder, self.query_encode_dim,
self.title_encode_dim, self.emb_size, self.emb_dim,
self.hidden_size, self.margin, self.query_len, self.pos_len,
self.neg_len)
cos_pos, cos_neg = simnet_model(inputs, is_infer)
self.inference_target_var = cos_pos
if is_infer:
fetch_dict = {'query_pt_sim': cos_pos}
return fetch_dict
loss_part1 = paddle.subtract(
paddle.full(shape=[self.batch_size, 1],
fill_value=self.margin,
dtype='float32'), cos_pos)
loss_part2 = paddle.add(loss_part1, cos_neg)
loss_part3 = paddle.maximum(
paddle.full(shape=[self.batch_size, 1],
fill_value=0.0,
dtype='float32'), loss_part2)
avg_cost = paddle.mean(loss_part3)
self._cost = avg_cost
self.acc = self.get_acc(cos_neg, cos_pos)
fetch_dict = {'Acc': self.acc, 'Loss': avg_cost}
return fetch_dict
def forward(self,
input_ids=None,
token_type_ids=None,
position_ids=None,
attention_mask=None,
query_input_ids=None,
query_token_type_ids=None,
query_position_ids=None,
query_attention_mask=None,
title_input_ids=None,
title_token_type_ids=None,
title_position_ids=None,
title_attention_mask=None,
seq_lengths=None,
labels=None):
if self.task != 'text-matching':
result = self.model(input_ids, token_type_ids, position_ids,
attention_mask)
else:
query_result = self.model(query_input_ids, query_token_type_ids,
query_position_ids, query_attention_mask)
title_result = self.model(title_input_ids, title_token_type_ids,
title_position_ids, title_attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(num_infer_chunks.numpy(),
num_label_chunks.numpy(),
num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
elif self.task == 'text-matching':
query_token_embedding = query_result
query_token_embedding = self.dropout(query_token_embedding)
query_attention_mask = paddle.unsqueeze(
(query_input_ids != self.model.pad_token_id).astype(
query_token_embedding.dtype),
axis=2)
query_token_embedding = query_token_embedding * query_attention_mask
query_sum_embedding = paddle.sum(query_token_embedding, axis=1)
query_sum_mask = paddle.sum(query_attention_mask, axis=1)
query_mean = query_sum_embedding / query_sum_mask
title_token_embedding = title_result
title_token_embedding = self.dropout(title_token_embedding)
title_attention_mask = paddle.unsqueeze(
(title_input_ids != self.model.pad_token_id).astype(
title_token_embedding.dtype),
axis=2)
title_token_embedding = title_token_embedding * title_attention_mask
title_sum_embedding = paddle.sum(title_token_embedding, axis=1)
title_sum_mask = paddle.sum(title_attention_mask, axis=1)
title_mean = title_sum_embedding / title_sum_mask
sub = paddle.abs(paddle.subtract(query_mean, title_mean))
projection = paddle.concat([query_mean, title_mean, sub], axis=-1)
logits = self.classifier(projection)
probs = F.softmax(logits)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
def test_quant_subtract(self):
out_1 = paddle.subtract(self.x, self.y)
out_2 = paddle.nn.quant.subtract()(self.x, self.y)
self.check(out_1, out_2)
def sub_prim2orig(op, x, y):
return paddle.subtract(x, y)
def add_subtract(inputs):
return paddle.subtract(paddle.add(inputs, inputs), inputs)
def subtract_wrapper(self, x):
return paddle.subtract(x[0], x[1])
def _executed_api(self, x, y, name=None):
return paddle.subtract(x, y, name)
def forward(self, x, y, name=None):
return paddle.subtract(x, y, name)
