Zobrazeno 1 - 10
of 83
pro vyhledávání: '"Ashish Sabharwal"'
Autor:
William Merrill, Ashish Sabharwal
Publikováno v:
Transactions of the Association for Computational Linguistics, Vol 11, Pp 531-545 (2023)
AbstractDespite their omnipresence in modern NLP, characterizing the computational power of transformer neural nets remains an interesting open question. We prove that transformers whose arithmetic precision is logarithmic in the number of input toke
Externí odkaz:
https://doaj.org/article/29f71320a4e74b8c80c7db0491d6dd2c
Publikováno v:
Proceedings of the International Conference on Automated Planning and Scheduling. 20:242-245
Upper Confidence bounds applied to Trees (UCT), a bandit-based Monte-Carlo sampling algorithm for planning, has recently been the subject of great interest in adversarial reasoning. UCT has been shown to outperform traditional minimax based approache
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::0c0fb889b43657def2ce35c0f9f73632
https://doi.org/10.1609/icaps.v20i1.13437
https://doi.org/10.1609/icaps.v20i1.13437
Autor:
Peter Clark, Oren Etzioni, Tushar Khot, Daniel Khashabi, Bhavana Mishra, Kyle Richardson, Ashish Sabharwal, Carissa Schoenick, Oyvind Tafjord, Niket Tandon, Sumithra Bhakthavatsalam, Dirk Groeneveld, Michal Guerquin, Michael Schmitz
Publikováno v:
AI Magazine; Vol. 41 No. 4: Winter 2020; 39-53
AI has achieved remarkable mastery over games such as Chess, Go, and Poker, and even Jeopardy!, but the rich variety of standardized exams has remained a landmark challenge. Even as recently as 2016, the best AI system could achieve merely 59.3 perce
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::12cbde13be63af9598aada27d315ba3f
https://doi.org/10.1609/aimag.v41i4.5304
https://doi.org/10.1609/aimag.v41i4.5304
Publikováno v:
AAAI
Composing knowledge from multiple pieces of texts is a key challenge in multi-hop question answering. We present a multi-hop reasoning dataset, Question Answering via Sentence Composition(QASC), that requires retrieving facts from a large corpus and
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::8bccd4681f9e433ee749a5e55fef2aa8
https://doi.org/10.1609/aaai.v34i05.6319
https://doi.org/10.1609/aaai.v34i05.6319
Training giant models from scratch for each complex task is resource- and data-inefficient. To help develop models that can leverage existing systems, we propose a new challenge: Learning to solve complex tasks by communicating with existing agents (
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::4700c90bd474b4839aa0b670cd575e5c
http://arxiv.org/abs/2110.08542
http://arxiv.org/abs/2110.08542
Multihop reasoning remains an elusive goal as existing multihop benchmarks are known to be largely solvable via shortcuts. Can we create a question answering (QA) dataset that, by construction, \emph{requires} proper multihop reasoning? To this end,
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::dfb233cc1441e1f121d7ce6723d58c5d
http://arxiv.org/abs/2108.00573
http://arxiv.org/abs/2108.00573
Publikováno v:
Frontiers in Artificial Intelligence and Applications
Model counting, or counting the number of solutions of a propositional formula, generalizes SAT and is the canonical #P-complete problem. Surprisingly, model counting is hard even for some polynomial-time solvable cases like 2-SAT and Horn-SAT. Effic
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::b1be676324a017ac344c2b1249daac1f
https://doi.org/10.3233/faia201009
https://doi.org/10.3233/faia201009
Research on incomplete algorithms for satisfiability testing lead to some of the first scalable SAT solvers in the early 1990’s. Unlike systematic solvers often based on an exhaustive branching and backtracking search, incomplete methods are genera
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::35b3139c6fa3da2351a314795ee053ca
https://doi.org/10.3233/faia200989
https://doi.org/10.3233/faia200989
Publikováno v:
ACL/IJCNLP (Findings)
Is it possible to use natural language to intervene in a model's behavior and alter its prediction in a desired way? We investigate the effectiveness of natural language interventions for reading-comprehension systems, studying this in the context of
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::0ae9b9fc2ab2eab22f0c88683edd0147
https://doi.org/10.18653/v1/2021.findings-acl.364
https://doi.org/10.18653/v1/2021.findings-acl.364
Publikováno v:
2021 Conference on Empirical Methods in Natural Language Processing-Proceedings of the Conference
Many real-world problems require the combined application of multiple reasoning abilities employing suitable abstractions, commonsense knowledge, and creative synthesis of problem-solving strategies. To help advance AI systems towards such capabiliti
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::985b2822593cde4b61da5f5635670215