Matches in SemOpenAlex for { <https://semopenalex.org/work/W3035068109> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W3035068109 abstract "Generative dialogue models currently suffer from a number of problems which standard maximum likelihood training does not address. They tend to produce generations that (i) rely too much on copying from the context, (ii) contain repetitions within utterances, (iii) overuse frequent words, and (iv) at a deeper level, contain logical flaws.In this work we show how all of these problems can be addressed by extending the recently introduced unlikelihood loss (Welleck et al., 2019) to these cases. We show that appropriate loss functions which regularize generated outputs to match human distributions are effective for the first three issues. For the last important general issue, we show applying unlikelihood to collected data of what a model should not do is effective for improving logical consistency, potentially paving the way to generative models with greater reasoning ability. We demonstrate the efficacy of our approach across several dialogue tasks." @default.
- W3035068109 created "2020-06-19" @default.
- W3035068109 creator A5001224830 @default.
- W3035068109 creator A5019030424 @default.
- W3035068109 creator A5049992350 @default.
- W3035068109 creator A5064261515 @default.
- W3035068109 creator A5070477596 @default.
- W3035068109 creator A5076635608 @default.
- W3035068109 creator A5091175785 @default.
- W3035068109 date "2020-01-01" @default.
- W3035068109 modified "2023-10-18" @default.
- W3035068109 title "Don’t Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training" @default.
- W3035068109 cites W1840435438 @default.
- W3035068109 cites W2741986794 @default.
- W3035068109 cites W2913443447 @default.
- W3035068109 cites W2916772188 @default.
- W3035068109 cites W2950681488 @default.
- W3035068109 cites W2955315729 @default.
- W3035068109 cites W2962788902 @default.
- W3035068109 cites W2962805889 @default.
- W3035068109 cites W2962974452 @default.
- W3035068109 cites W2962989446 @default.
- W3035068109 cites W2963149412 @default.
- W3035068109 cites W2963403868 @default.
- W3035068109 cites W2963475460 @default.
- W3035068109 cites W2963531095 @default.
- W3035068109 cites W2963825865 @default.
- W3035068109 cites W2963846996 @default.
- W3035068109 cites W2969574947 @default.
- W3035068109 cites W2970453125 @default.
- W3035068109 cites W2971883198 @default.
- W3035068109 cites W2995404354 @default.
- W3035068109 cites W2996287690 @default.
- W3035068109 cites W2997012196 @default.
- W3035068109 cites W3035239386 @default.
- W3035068109 doi "https://doi.org/10.18653/v1/2020.acl-main.428" @default.
- W3035068109 hasPublicationYear "2020" @default.
- W3035068109 type Work @default.
- W3035068109 sameAs 3035068109 @default.
- W3035068109 citedByCount "71" @default.
- W3035068109 countsByYear W30350681092020 @default.
- W3035068109 countsByYear W30350681092021 @default.
- W3035068109 countsByYear W30350681092022 @default.
- W3035068109 countsByYear W30350681092023 @default.
- W3035068109 crossrefType "proceedings-article" @default.
- W3035068109 hasAuthorship W3035068109A5001224830 @default.
- W3035068109 hasAuthorship W3035068109A5019030424 @default.
- W3035068109 hasAuthorship W3035068109A5049992350 @default.
- W3035068109 hasAuthorship W3035068109A5064261515 @default.
- W3035068109 hasAuthorship W3035068109A5070477596 @default.
- W3035068109 hasAuthorship W3035068109A5076635608 @default.
- W3035068109 hasAuthorship W3035068109A5091175785 @default.
- W3035068109 hasBestOaLocation W30350681091 @default.
- W3035068109 hasConcept C119857082 @default.
- W3035068109 hasConcept C151730666 @default.
- W3035068109 hasConcept C154945302 @default.
- W3035068109 hasConcept C167966045 @default.
- W3035068109 hasConcept C2522767166 @default.
- W3035068109 hasConcept C2776436953 @default.
- W3035068109 hasConcept C2779343474 @default.
- W3035068109 hasConcept C39890363 @default.
- W3035068109 hasConcept C41008148 @default.
- W3035068109 hasConcept C86803240 @default.
- W3035068109 hasConceptScore W3035068109C119857082 @default.
- W3035068109 hasConceptScore W3035068109C151730666 @default.
- W3035068109 hasConceptScore W3035068109C154945302 @default.
- W3035068109 hasConceptScore W3035068109C167966045 @default.
- W3035068109 hasConceptScore W3035068109C2522767166 @default.
- W3035068109 hasConceptScore W3035068109C2776436953 @default.
- W3035068109 hasConceptScore W3035068109C2779343474 @default.
- W3035068109 hasConceptScore W3035068109C39890363 @default.
- W3035068109 hasConceptScore W3035068109C41008148 @default.
- W3035068109 hasConceptScore W3035068109C86803240 @default.
- W3035068109 hasLocation W30350681091 @default.
- W3035068109 hasLocation W30350681092 @default.
- W3035068109 hasOpenAccess W3035068109 @default.
- W3035068109 hasPrimaryLocation W30350681091 @default.
- W3035068109 hasRelatedWork W1534961803 @default.
- W3035068109 hasRelatedWork W1576360539 @default.
- W3035068109 hasRelatedWork W2770426046 @default.
- W3035068109 hasRelatedWork W3111180383 @default.
- W3035068109 hasRelatedWork W3165625257 @default.
- W3035068109 hasRelatedWork W4285020345 @default.
- W3035068109 hasRelatedWork W4299531873 @default.
- W3035068109 hasRelatedWork W4313532828 @default.
- W3035068109 hasRelatedWork W4385573542 @default.
- W3035068109 hasRelatedWork W2310403681 @default.
- W3035068109 isParatext "false" @default.
- W3035068109 isRetracted "false" @default.
- W3035068109 magId "3035068109" @default.
- W3035068109 workType "article" @default.