SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Model output test (#6155)

Browse Source 
* Use return_dict=True in all tests

* Formatting
This commit is contained in:
Sylvain Gugger
2020-07-31 09:44:37 -04:00
committed by GitHub
parent 86caab1e0b
commit d951c14ae4
26 changed files with 320 additions and 765 deletions
Download Patch File Download Diff File Expand all files Collapse all files

75
tests/test_modeling_longformer.py
View File

@@ -108,6 +108,7 @@ class LongformerModelTester:
type_vocab_size=self.type_vocab_size,
initializer_range=self.initializer_range,
attention_window=self.attention_window,
return_dict=True,
)
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
@@ -123,8 +124,8 @@ class LongformerModelTester:
model.eval()
attention_mask = torch.ones(input_ids.shape, dtype=torch.long, device=torch_device)
output_with_mask = model(input_ids, attention_mask=attention_mask)[0]
output_without_mask = model(input_ids)[0]
output_with_mask = model(input_ids, attention_mask=attention_mask)["last_hidden_state"]
output_without_mask = model(input_ids)["last_hidden_state"]
self.parent.assertTrue(torch.allclose(output_with_mask[0, 0, :5], output_without_mask[0, 0, :5], atol=1e-4))
def create_and_check_longformer_model(
@@ -133,18 +134,13 @@ class LongformerModelTester:
model = LongformerModel(config=config)
model.to(torch_device)
model.eval()
sequence_output, pooled_output = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids)
sequence_output, pooled_output = model(input_ids, token_type_ids=token_type_ids)
sequence_output, pooled_output = model(input_ids)
result = {
"sequence_output": sequence_output,
"pooled_output": pooled_output,
}
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids)
result = model(input_ids, token_type_ids=token_type_ids)
result = model(input_ids)
self.parent.assertListEqual(
list(result["sequence_output"].size()), [self.batch_size, self.seq_length, self.hidden_size]
list(result["last_hidden_state"].size()), [self.batch_size, self.seq_length, self.hidden_size]
)
self.parent.assertListEqual(list(result["pooled_output"].size()), [self.batch_size, self.hidden_size])
self.parent.assertListEqual(list(result["pooler_output"].size()), [self.batch_size, self.hidden_size])
def create_and_check_longformer_model_with_global_attention_mask(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
@@ -156,25 +152,19 @@ class LongformerModelTester:
global_attention_mask[:, input_mask.shape[-1] // 2] = 0
global_attention_mask = global_attention_mask.to(torch_device)
sequence_output, pooled_output = model(
result = model(
input_ids,
attention_mask=input_mask,
global_attention_mask=global_attention_mask,
token_type_ids=token_type_ids,
)
sequence_output, pooled_output = model(
input_ids, token_type_ids=token_type_ids, global_attention_mask=global_attention_mask
)
sequence_output, pooled_output = model(input_ids, global_attention_mask=global_attention_mask)
result = model(input_ids, token_type_ids=token_type_ids, global_attention_mask=global_attention_mask)
result = model(input_ids, global_attention_mask=global_attention_mask)
result = {
"sequence_output": sequence_output,
"pooled_output": pooled_output,
}
self.parent.assertListEqual(
list(result["sequence_output"].size()), [self.batch_size, self.seq_length, self.hidden_size]
list(result["last_hidden_state"].size()), [self.batch_size, self.seq_length, self.hidden_size]
)
self.parent.assertListEqual(list(result["pooled_output"].size()), [self.batch_size, self.hidden_size])
self.parent.assertListEqual(list(result["pooler_output"].size()), [self.batch_size, self.hidden_size])
def create_and_check_longformer_for_masked_lm(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
@@ -182,16 +172,8 @@ class LongformerModelTester:
model = LongformerForMaskedLM(config=config)
model.to(torch_device)
model.eval()
loss, prediction_scores = model(
input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels
)
result = {
"loss": loss,
"prediction_scores": prediction_scores,
}
self.parent.assertListEqual(
list(result["prediction_scores"].size()), [self.batch_size, self.seq_length, self.vocab_size]
)
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
self.parent.assertListEqual(list(result["logits"].size()), [self.batch_size, self.seq_length, self.vocab_size])
self.check_loss_output(result)
def create_and_check_longformer_for_question_answering(
@@ -200,7 +182,7 @@ class LongformerModelTester:
model = LongformerForQuestionAnswering(config=config)
model.to(torch_device)
model.eval()
loss, start_logits, end_logits = model(
result = model(
input_ids,
attention_mask=input_mask,
global_attention_mask=input_mask,
@@ -208,11 +190,6 @@ class LongformerModelTester:
start_positions=sequence_labels,
end_positions=sequence_labels,
)
result = {
"loss": loss,
"start_logits": start_logits,
"end_logits": end_logits,
}
self.parent.assertListEqual(list(result["start_logits"].size()), [self.batch_size, self.seq_length])
self.parent.assertListEqual(list(result["end_logits"].size()), [self.batch_size, self.seq_length])
self.check_loss_output(result)
@@ -224,13 +201,7 @@ class LongformerModelTester:
model = LongformerForSequenceClassification(config)
model.to(torch_device)
model.eval()
loss, logits = model(
input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=sequence_labels
)
result = {
"loss": loss,
"logits": logits,
}
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=sequence_labels)
self.parent.assertListEqual(list(result["logits"].size()), [self.batch_size, self.num_labels])
self.check_loss_output(result)
@@ -241,11 +212,7 @@ class LongformerModelTester:
model = LongformerForTokenClassification(config=config)
model.to(torch_device)
model.eval()
loss, logits = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
result = {
"loss": loss,
"logits": logits,
}
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
self.parent.assertListEqual(list(result["logits"].size()), [self.batch_size, self.seq_length, self.num_labels])
self.check_loss_output(result)
@@ -260,17 +227,13 @@ class LongformerModelTester:
multiple_choice_token_type_ids = token_type_ids.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
multiple_choice_input_mask = input_mask.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
multiple_choice_input_mask = input_mask.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
loss, logits = model(
result = model(
multiple_choice_inputs_ids,
attention_mask=multiple_choice_input_mask,
global_attention_mask=multiple_choice_input_mask,
token_type_ids=multiple_choice_token_type_ids,
labels=choice_labels,
)
result = {
"loss": loss,
"logits": logits,
}
self.parent.assertListEqual(list(result["logits"].size()), [self.batch_size, self.num_choices])
self.check_loss_output(result)