XNLI: Natural Language Inference
LCQMC: Semantic similarity task, input two sentences, output whether the two sentences are similar.
LCQMC data set format:
MSRA-NER: Chinese Named Entity Recognition
ChnSentiCorp: Chinese Sentiment Analysis
Nlpcc-dbqa: Retrieval Question Answering Matching Task
SQuAD (Standford Question Answering Dataset) task: question and answer matching task
MNLI (Multi-Genre Natural Language Inference): Judging sentences are similar, contradictory or irrelevant
QNLI: Question-answering NLI: Natural Language Reasoning, formerly SQuAD 1.0
QQP (Quora Question Pairs): Determine whether two Quora questions are equivalent
RTE (Recognizing Textual Entailment): Similar to MNLI, but with a smaller amount of data
QNLI (Question Natural Language Inference): Determine whether a pair of QA is corresponding
SST-2 (Stanford Sentiment Treebank): Sentiment Classification
CoLA (Corpus of Linguistic Acceptability): Judging whether an English sentence is grammatically correct
MRPC (Microsoft Research Paraphrase Corpus): Determine whether the semantics of two comments are the same
CoLA (Corpus of Linguistic Acceptability): A two-category problem for a single sentence, judging whether an English sentence is grammatically acceptable.
STS-B (Semantic Textual Similarity Benchmark): Calculate the similarity of sentence pairs
NLI: (Natural Language Inference), judging that two sentences have the same semantics, neutrality, and opposition, three classification tasks
New state-of-the-art results: MNLI, QNLI, RTE and STS-B.
MNLI (Multi-Genre Natural Language Inference): Judging sentences are similar, contradictory or irrelevant
QNLI (Question Natural Language Inference): Determine whether a pair of QA is corresponding
RTE (Recognizing Textual Entailment): Similar to MNLI, but with a smaller amount of data
STS-B (Semantic Textual Similarity Benchmark): Calculate the similarity of sentence pairs
Applications of text matching tasks:
Search, recommendation, Q&A, etc.