@article{efrat_cryptonite_2021,
	title = {Cryptonite: {A} {Cryptic} {Crossword} {Benchmark} for {Extreme} {Ambiguity} in {Language}},
	shorttitle = {Cryptonite},
	url = {http://arxiv.org/abs/2103.01242},
	abstract = {Current NLP datasets targeting ambiguity can be solved by a native speaker with relative ease. We present Cryptonite, a large-scale dataset based on cryptic crosswords, which is both linguistically complex and naturally sourced. Each example in Cryptonite is a cryptic clue, a short phrase or sentence with a misleading surface reading, whose solving requires disambiguating semantic, syntactic, and phonetic wordplays, as well as world knowledge. Cryptic clues pose a challenge even for experienced solvers, though top-tier experts can solve them with almost 100\% accuracy. Cryptonite is a challenging task for current models; fine-tuning T5-Large on 470k cryptic clues achieves only 7.6\% accuracy, on par with the accuracy of a rule-based clue solver (8.6\%).},
	urldate = {2022-02-14},
	journal = {arXiv:2103.01242 [cs, stat]},
	author = {Efrat, Avia and Shaham, Uri and Kilman, Dan and Levy, Omer},
	month = nov,
	year = {2021},
	note = {arXiv: 2103.01242},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Machine Learning, Statistics - Machine Learning},
}

{"_id":"X8QrtQXJYoTqfJMPb","bibbaseid":"efrat-shaham-kilman-levy-cryptoniteacrypticcrosswordbenchmarkforextremeambiguityinlanguage-2021","author_short":["Efrat, A.","Shaham, U.","Kilman, D.","Levy, O."],"bibdata":{"bibtype":"article","type":"article","title":"Cryptonite: A Cryptic Crossword Benchmark for Extreme Ambiguity in Language","shorttitle":"Cryptonite","url":"http://arxiv.org/abs/2103.01242","abstract":"Current NLP datasets targeting ambiguity can be solved by a native speaker with relative ease. We present Cryptonite, a large-scale dataset based on cryptic crosswords, which is both linguistically complex and naturally sourced. Each example in Cryptonite is a cryptic clue, a short phrase or sentence with a misleading surface reading, whose solving requires disambiguating semantic, syntactic, and phonetic wordplays, as well as world knowledge. Cryptic clues pose a challenge even for experienced solvers, though top-tier experts can solve them with almost 100% accuracy. Cryptonite is a challenging task for current models; fine-tuning T5-Large on 470k cryptic clues achieves only 7.6% accuracy, on par with the accuracy of a rule-based clue solver (8.6%).","urldate":"2022-02-14","journal":"arXiv:2103.01242 [cs, stat]","author":[{"propositions":[],"lastnames":["Efrat"],"firstnames":["Avia"],"suffixes":[]},{"propositions":[],"lastnames":["Shaham"],"firstnames":["Uri"],"suffixes":[]},{"propositions":[],"lastnames":["Kilman"],"firstnames":["Dan"],"suffixes":[]},{"propositions":[],"lastnames":["Levy"],"firstnames":["Omer"],"suffixes":[]}],"month":"November","year":"2021","note":"arXiv: 2103.01242","keywords":"Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Machine Learning, Statistics - Machine Learning","bibtex":"@article{efrat_cryptonite_2021,\n\ttitle = {Cryptonite: {A} {Cryptic} {Crossword} {Benchmark} for {Extreme} {Ambiguity} in {Language}},\n\tshorttitle = {Cryptonite},\n\turl = {http://arxiv.org/abs/2103.01242},\n\tabstract = {Current NLP datasets targeting ambiguity can be solved by a native speaker with relative ease. We present Cryptonite, a large-scale dataset based on cryptic crosswords, which is both linguistically complex and naturally sourced. Each example in Cryptonite is a cryptic clue, a short phrase or sentence with a misleading surface reading, whose solving requires disambiguating semantic, syntactic, and phonetic wordplays, as well as world knowledge. Cryptic clues pose a challenge even for experienced solvers, though top-tier experts can solve them with almost 100\\% accuracy. Cryptonite is a challenging task for current models; fine-tuning T5-Large on 470k cryptic clues achieves only 7.6\\% accuracy, on par with the accuracy of a rule-based clue solver (8.6\\%).},\n\turldate = {2022-02-14},\n\tjournal = {arXiv:2103.01242 [cs, stat]},\n\tauthor = {Efrat, Avia and Shaham, Uri and Kilman, Dan and Levy, Omer},\n\tmonth = nov,\n\tyear = {2021},\n\tnote = {arXiv: 2103.01242},\n\tkeywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Machine Learning, Statistics - Machine Learning},\n}\n\n","author_short":["Efrat, A.","Shaham, U.","Kilman, D.","Levy, O."],"key":"efrat_cryptonite_2021","id":"efrat_cryptonite_2021","bibbaseid":"efrat-shaham-kilman-levy-cryptoniteacrypticcrosswordbenchmarkforextremeambiguityinlanguage-2021","role":"author","urls":{"Paper":"http://arxiv.org/abs/2103.01242"},"keyword":["Computer Science - Artificial Intelligence","Computer Science - Computation and Language","Computer Science - Machine Learning","Statistics - Machine Learning"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/mxmplx","dataSources":["nTukRxsNNmbKgjGgn","aXmRAq63YsH7a3ufx"],"keywords":["computer science - artificial intelligence","computer science - computation and language","computer science - machine learning","statistics - machine learning"],"search_terms":["cryptonite","cryptic","crossword","benchmark","extreme","ambiguity","language","efrat","shaham","kilman","levy"],"title":"Cryptonite: A Cryptic Crossword Benchmark for Extreme Ambiguity in Language","year":2021}