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

Black preview (#17217)

Browse Source 
* Black preview

* Fixup too!

* Fix check copies

* Use the same version as the CI

* Bump black
This commit is contained in:
Sylvain Gugger
2022-05-12 16:25:55 -04:00
committed by GitHub
parent 9bd67ac7bb
commit afe5d42d8d
578 changed files with 8274 additions and 3296 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/pipelines/test_pipelines_automatic_speech_recognition.py
View File

@@ -184,7 +184,9 @@ class AutomaticSpeechRecognitionPipelineTests(unittest.TestCase, metaclass=Pipel
self.assertEqual(
output,
{
"text": "y en las ramas medio sumergidas revoloteaban algunos pájaros de quimérico y legendario plumajre"
"text": (
"y en las ramas medio sumergidas revoloteaban algunos pájaros de quimérico y legendario plumajre"
)
},
)
@@ -194,7 +196,9 @@ class AutomaticSpeechRecognitionPipelineTests(unittest.TestCase, metaclass=Pipel
self.assertEqual(
output,
{
"text": "y en las ramas medio sumergidas revoloteaban algunos pájaros de quimérico y legendario plumajcri",
"text": (
"y en las ramas medio sumergidas revoloteaban algunos pájaros de quimérico y legendario plumajcri"
),
"chunks": [
{"text": "y", "timestamp": (0.52, 0.54)},
{"text": "en", "timestamp": (0.6, 0.68)},

3
tests/pipelines/test_pipelines_common.py
View File

@@ -184,7 +184,8 @@ class PipelineTestCaseMeta(type):
if tokenizer is None and feature_extractor is None:
self.skipTest(
f"Ignoring {ModelClass}, cannot create a tokenizer or feature_extractor (PerceiverConfig with no FastTokenizer ?)"
f"Ignoring {ModelClass}, cannot create a tokenizer or feature_extractor (PerceiverConfig with"
" no FastTokenizer ?)"
)
pipeline, examples = self.get_test_pipeline(model, tokenizer, feature_extractor)
if pipeline is None:

37
tests/pipelines/test_pipelines_question_answering.py
View File

@@ -199,7 +199,42 @@ class QAPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
)
outputs = qa_pipeline(
{
"context": "Yes Bank founder Rana Kapoor has approached the Bombay High Court, challenging a special court's order from August this year that had remanded him in police custody for a week in a multi-crore loan fraud case. Kapoor, who is currently lodged in Taloja Jail, is an accused in the loan fraud case and some related matters being probed by the CBI and Enforcement Directorate. A single bench presided over by Justice S K Shinde on Tuesday posted the plea for further hearing on October 14. In his plea filed through advocate Vijay Agarwal, Kapoor claimed that the special court's order permitting the CBI's request for police custody on August 14 was illegal and in breach of the due process of law. Therefore, his police custody and subsequent judicial custody in the case were all illegal. Kapoor has urged the High Court to quash and set aside the special court's order dated August 14. As per his plea, in August this year, the CBI had moved two applications before the special court, one seeking permission to arrest Kapoor, who was already in judicial custody at the time in another case, and the other, seeking his police custody. While the special court refused to grant permission to the CBI to arrest Kapoor, it granted the central agency's plea for his custody. Kapoor, however, said in his plea that before filing an application for his arrest, the CBI had not followed the process of issuing him a notice under Section 41 of the CrPC for appearance before it. He further said that the CBI had not taken prior sanction as mandated under section 17 A of the Prevention of Corruption Act for prosecuting him. The special court, however, had said in its order at the time that as Kapoor was already in judicial custody in another case and was not a free man the procedure mandated under Section 41 of the CrPC need not have been adhered to as far as issuing a prior notice of appearance was concerned. ADVERTISING It had also said that case records showed that the investigating officer had taken an approval from a managing director of Yes Bank before beginning the proceedings against Kapoor and such a permission was a valid sanction. However, Kapoor in his plea said that the above order was bad in law and sought that it be quashed and set aside. The law mandated that if initial action was not in consonance with legal procedures, then all subsequent actions must be held as illegal, he said, urging the High Court to declare the CBI remand and custody and all subsequent proceedings including the further custody as illegal and void ab-initio. In a separate plea before the High Court, Kapoor's daughter Rakhee Kapoor-Tandon has sought exemption from in-person appearance before a special PMLA court. Rakhee has stated that she is a resident of the United Kingdom and is unable to travel to India owing to restrictions imposed due to the COVID-19 pandemic. According to the CBI, in the present case, Kapoor had obtained a gratification or pecuniary advantage of ₹ 307 crore, and thereby caused Yes Bank a loss of ₹ 1,800 crore by extending credit facilities to Avantha Group, when it was not eligible for the same",
"context": (
"Yes Bank founder Rana Kapoor has approached the Bombay High Court, challenging a special court's"
" order from August this year that had remanded him in police custody for a week in a multi-crore"
" loan fraud case. Kapoor, who is currently lodged in Taloja Jail, is an accused in the loan fraud"
" case and some related matters being probed by the CBI and Enforcement Directorate. A single"
" bench presided over by Justice S K Shinde on Tuesday posted the plea for further hearing on"
" October 14. In his plea filed through advocate Vijay Agarwal, Kapoor claimed that the special"
" court's order permitting the CBI's request for police custody on August 14 was illegal and in"
" breach of the due process of law. Therefore, his police custody and subsequent judicial custody"
" in the case were all illegal. Kapoor has urged the High Court to quash and set aside the special"
" court's order dated August 14. As per his plea, in August this year, the CBI had moved two"
" applications before the special court, one seeking permission to arrest Kapoor, who was already"
" in judicial custody at the time in another case, and the other, seeking his police custody."
" While the special court refused to grant permission to the CBI to arrest Kapoor, it granted the"
" central agency's plea for his custody. Kapoor, however, said in his plea that before filing an"
" application for his arrest, the CBI had not followed the process of issuing him a notice under"
" Section 41 of the CrPC for appearance before it. He further said that the CBI had not taken"
" prior sanction as mandated under section 17 A of the Prevention of Corruption Act for"
" prosecuting him. The special court, however, had said in its order at the time that as Kapoor"
" was already in judicial custody in another case and was not a free man the procedure mandated"
" under Section 41 of the CrPC need not have been adhered to as far as issuing a prior notice of"
" appearance was concerned. ADVERTISING It had also said that case records showed that the"
" investigating officer had taken an approval from a managing director of Yes Bank before"
" beginning the proceedings against Kapoor and such a permission was a valid sanction. However,"
" Kapoor in his plea said that the above order was bad in law and sought that it be quashed and"
" set aside. The law mandated that if initial action was not in consonance with legal procedures,"
" then all subsequent actions must be held as illegal, he said, urging the High Court to declare"
" the CBI remand and custody and all subsequent proceedings including the further custody as"
" illegal and void ab-initio. In a separate plea before the High Court, Kapoor's daughter Rakhee"
" Kapoor-Tandon has sought exemption from in-person appearance before a special PMLA court. Rakhee"
" has stated that she is a resident of the United Kingdom and is unable to travel to India owing"
" to restrictions imposed due to the COVID-19 pandemic. According to the CBI, in the present case,"
" Kapoor had obtained a gratification or pecuniary advantage of ₹ 307 crore, and thereby caused"
" Yes Bank a loss of ₹ 1,800 crore by extending credit facilities to Avantha Group, when it was"
" not eligible for the same"
),
"question": "Is this person invovled in fraud?",
}
)

46
tests/pipelines/test_pipelines_summarization.py
View File

@@ -91,7 +91,49 @@ class SummarizationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMe
@slow
def test_integration_torch_summarization(self):
summarizer = pipeline(task="summarization", device=DEFAULT_DEVICE_NUM)
cnn_article = ' (CNN)The Palestinian Authority officially became the 123rd member of the International Criminal Court on Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based. The Palestinians signed the ICC\'s founding Rome Statute in January, when they also accepted its jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the situation in Palestinian territories, paving the way for possible war crimes investigations against Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and the United States, neither of which is an ICC member, opposed the Palestinians\' efforts to join the body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday\'s ceremony, said it was a move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the world is also a step closer to ending a long era of impunity and injustice," he said, according to an ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine acquires all the rights as well as responsibilities that come with being a State Party to the Statute. These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should immediately end their pressure, and countries that support universal acceptance of the court\'s treaty should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the group. "What\'s objectionable is the attempts to undermine international justice, not Palestine\'s decision to join a treaty to which over 100 countries around the world are members." In January, when the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an outrage, saying the court was overstepping its boundaries. The United States also said it "strongly" disagreed with the court\'s decision. "As we have said repeatedly, we do not believe that Palestine is a state and therefore we do not believe that it is eligible to join the ICC," the State Department said in a statement. It urged the warring sides to resolve their differences through direct negotiations. "We will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace," it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou Bensouda said her office would "conduct its analysis in full independence and impartiality." The war between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry will include alleged war crimes committed since June. The International Criminal Court was set up in 2002 to prosecute genocide, crimes against humanity and war crimes. CNN\'s Vasco Cotovio, Kareem Khadder and Faith Karimi contributed to this report.'
expected_cnn_summary = " The Palestinian Authority becomes the 123rd member of the International Criminal Court . The move gives the court jurisdiction over alleged crimes in Palestinian territories . Israel and the United States opposed the Palestinians' efforts to join the court . Rights group Human Rights Watch welcomes the move, says governments seeking to penalize Palestine should end pressure ."
cnn_article = (
" (CNN)The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes. CNN's Vasco Cotovio, Kareem Khadder"
" and Faith Karimi contributed to this report."
)
expected_cnn_summary = (
" The Palestinian Authority becomes the 123rd member of the International Criminal Court . The move gives"
" the court jurisdiction over alleged crimes in Palestinian territories . Israel and the United States"
" opposed the Palestinians' efforts to join the court . Rights group Human Rights Watch welcomes the move,"
" says governments seeking to penalize Palestine should end pressure ."
)
result = summarizer(cnn_article)
self.assertEqual(result[0]["summary_text"], expected_cnn_summary)

12
tests/pipelines/test_pipelines_table_question_answering.py
View File

@@ -92,7 +92,8 @@ class TQAPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
},
query=[
"What repository has the largest number of stars?",
"Given that the numbers of stars defines if a repository is active, what repository is the most active?",
"Given that the numbers of stars defines if a repository is active, what repository is the most"
" active?",
"What is the number of repositories?",
"What is the average number of stars?",
"What is the total amount of stars?",
@@ -194,7 +195,8 @@ class TQAPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
},
query=[
"What repository has the largest number of stars?",
"Given that the numbers of stars defines if a repository is active, what repository is the most active?",
"Given that the numbers of stars defines if a repository is active, what repository is the most"
" active?",
"What is the number of repositories?",
"What is the average number of stars?",
"What is the total amount of stars?",
@@ -313,7 +315,8 @@ class TQAPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
},
query=[
"What repository has the largest number of stars?",
"Given that the numbers of stars defines if a repository is active, what repository is the most active?",
"Given that the numbers of stars defines if a repository is active, what repository is the most"
" active?",
"What is the number of repositories?",
"What is the average number of stars?",
"What is the total amount of stars?",
@@ -434,7 +437,8 @@ class TQAPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
},
query=[
"What repository has the largest number of stars?",
"Given that the numbers of stars defines if a repository is active, what repository is the most active?",
"Given that the numbers of stars defines if a repository is active, what repository is the most"
" active?",
"What is the number of repositories?",
"What is the average number of stars?",
"What is the total amount of stars?",

30
tests/pipelines/test_pipelines_text_generation.py
View File

@@ -34,7 +34,10 @@ class TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseM
outputs,
[
{
"generated_text": "This is a test ☃ ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy oscope. oscope. FiliFili@@"
"generated_text": (
"This is a test ☃ ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy oscope."
" oscope. FiliFili@@"
)
}
],
)
@@ -45,12 +48,18 @@ class TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseM
[
[
{
"generated_text": "This is a test ☃ ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy oscope. oscope. FiliFili@@"
"generated_text": (
"This is a test ☃ ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy oscope."
" oscope. FiliFili@@"
)
}
],
[
{
"generated_text": "This is a second test ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy oscope. oscope. FiliFili@@"
"generated_text": (
"This is a second test ☃ segmental segmental segmental 议议eski eski flutter flutter Lacy"
" oscope. oscope. FiliFili@@"
)
}
],
],
@@ -97,7 +106,10 @@ class TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseM
outputs,
[
{
"generated_text": "This is a test FeyFeyFey(Croatis.), s.), Cannes Cannes Cannes 閲閲Cannes Cannes Cannes 攵 please,"
"generated_text": (
"This is a test FeyFeyFey(Croatis.), s.), Cannes Cannes Cannes 閲閲Cannes Cannes Cannes 攵"
" please,"
)
}
],
)
@@ -108,12 +120,18 @@ class TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseM
[
[
{
"generated_text": "This is a test FeyFeyFey(Croatis.), s.), Cannes Cannes Cannes 閲閲Cannes Cannes Cannes 攵 please,"
"generated_text": (
"This is a test FeyFeyFey(Croatis.), s.), Cannes Cannes Cannes 閲閲Cannes Cannes Cannes 攵"
" please,"
)
}
],
[
{
"generated_text": "This is a second test Chieftain Chieftain prefecture prefecture prefecture Cannes Cannes Cannes 閲閲Cannes Cannes Cannes 攵 please,"
"generated_text": (
"This is a second test Chieftain Chieftain prefecture prefecture prefecture Cannes Cannes"
" Cannes 閲閲Cannes Cannes Cannes 攵 please,"
)
}
],
],

20
tests/pipelines/test_pipelines_translation.py
View File

@@ -61,7 +61,10 @@ class TranslationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta
outputs,
[
{
"translation_text": "Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide"
"translation_text": (
"Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide"
" Beide Beide"
)
}
],
)
@@ -74,7 +77,10 @@ class TranslationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta
outputs,
[
{
"translation_text": "Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide"
"translation_text": (
"Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide Beide"
" Beide Beide"
)
}
],
)
@@ -87,7 +93,10 @@ class TranslationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta
outputs,
[
{
"translation_text": "monoton monoton monoton monoton monoton monoton monoton monoton monoton monoton urine urine urine urine urine urine urine urine urine"
"translation_text": (
"monoton monoton monoton monoton monoton monoton monoton monoton monoton monoton urine urine"
" urine urine urine urine urine urine urine"
)
}
],
)
@@ -100,7 +109,10 @@ class TranslationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta
outputs,
[
{
"translation_text": "monoton monoton monoton monoton monoton monoton monoton monoton monoton monoton urine urine urine urine urine urine urine urine urine"
"translation_text": (
"monoton monoton monoton monoton monoton monoton monoton monoton monoton monoton urine urine"
" urine urine urine urine urine urine urine"
)
}
],
)

58
tests/pipelines/test_pipelines_zero_shot.py
View File

@@ -202,14 +202,39 @@ class ZeroShotClassificationPipelineTests(unittest.TestCase, metaclass=PipelineT
},
)
outputs = zero_shot_classifier(
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.",
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks"
" in an encoder-decoder configuration. The best performing models also connect the encoder and decoder"
" through an attention mechanism. We propose a new simple network architecture, the Transformer, based"
" solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two"
" machine translation tasks show these models to be superior in quality while being more parallelizable"
" and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014"
" English-to-German translation task, improving over the existing best results, including ensembles by"
" over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new"
" single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small"
" fraction of the training costs of the best models from the literature. We show that the Transformer"
" generalizes well to other tasks by applying it successfully to English constituency parsing both with"
" large and limited training data.",
candidate_labels=["machine learning", "statistics", "translation", "vision"],
multi_label=True,
)
self.assertEqual(
nested_simplify(outputs),
{
"sequence": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.",
"sequence": (
"The dominant sequence transduction models are based on complex recurrent or convolutional neural"
" networks in an encoder-decoder configuration. The best performing models also connect the"
" encoder and decoder through an attention mechanism. We propose a new simple network"
" architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence"
" and convolutions entirely. Experiments on two machine translation tasks show these models to be"
" superior in quality while being more parallelizable and requiring significantly less time to"
" train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task,"
" improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014"
" English-to-French translation task, our model establishes a new single-model state-of-the-art"
" BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training"
" costs of the best models from the literature. We show that the Transformer generalizes well to"
" other tasks by applying it successfully to English constituency parsing both with large and"
" limited training data."
),
"labels": ["translation", "machine learning", "vision", "statistics"],
"scores": [0.817, 0.713, 0.018, 0.018],
},
@@ -232,14 +257,39 @@ class ZeroShotClassificationPipelineTests(unittest.TestCase, metaclass=PipelineT
},
)
outputs = zero_shot_classifier(
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.",
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks"
" in an encoder-decoder configuration. The best performing models also connect the encoder and decoder"
" through an attention mechanism. We propose a new simple network architecture, the Transformer, based"
" solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two"
" machine translation tasks show these models to be superior in quality while being more parallelizable"
" and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014"
" English-to-German translation task, improving over the existing best results, including ensembles by"
" over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new"
" single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small"
" fraction of the training costs of the best models from the literature. We show that the Transformer"
" generalizes well to other tasks by applying it successfully to English constituency parsing both with"
" large and limited training data.",
candidate_labels=["machine learning", "statistics", "translation", "vision"],
multi_label=True,
)
self.assertEqual(
nested_simplify(outputs),
{
"sequence": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.",
"sequence": (
"The dominant sequence transduction models are based on complex recurrent or convolutional neural"
" networks in an encoder-decoder configuration. The best performing models also connect the"
" encoder and decoder through an attention mechanism. We propose a new simple network"
" architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence"
" and convolutions entirely. Experiments on two machine translation tasks show these models to be"
" superior in quality while being more parallelizable and requiring significantly less time to"
" train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task,"
" improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014"
" English-to-French translation task, our model establishes a new single-model state-of-the-art"
" BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training"
" costs of the best models from the literature. We show that the Transformer generalizes well to"
" other tasks by applying it successfully to English constituency parsing both with large and"
" limited training data."
),
"labels": ["translation", "machine learning", "vision", "statistics"],
"scores": [0.817, 0.713, 0.018, 0.018],
},