We investigate evaluation metrics for dialogue response generation systems where supervised labels, such as task completion, are not available. Recent works in response generation have adopted metrics from machine translation to compare a model's generated response to a single target response. We show that these metrics correlate very weakly with human judgements in the non-technical Twitter domain, and not at all in the technical Ubuntu domain. We provide quantitative and qualitative results highlighting specific weaknesses in existing metrics, and provide recommendations for future development of better automatic evaluation metrics for dialogue systems.

No abstract

No abstract

Abstract:We study the semileptonic and non-leptonic charmed baryon decays with SU(3) flavor symmetry, where the charmed baryons can be B c = (n denoted as the baryon octet (decuplet), we find that the B c → B ′ n ℓ + ν ℓ decays are forbidden, while theand Ω ++ ccc → Ω + cc ℓ + ν ℓ decays are the only existing Cabibboallowed modes for B ′ c → B ′ n ℓ + ν ℓ , B cc → B ′ c ℓ + ν ℓ , and B ccc → B (′) cc ℓ + ν ℓ , respectively. We predict the rarely studied.3, 4.8±0.5)×10 −3 . For the observation, the doubly and triply charmed baryon decays ofare the favored Cabibboallowed decays, which are accessible to the BESIII and LHCb experiments.

No abstract

We explore the breaking effects of the SU (3) flavor symmetry in the singly Cabibbo-suppressed antitriplet charmed baryon decays of B c → B n M, with B c = ( 0 c , + c , + c ) and B n (M) the baryon (pseudo-scalar) octets. We find that these breaking effects can be used to account for the experimental data on the decay branching ratios of B(In addition, we obtain that, which all receive significant contributions from the breaking effects, and can be tested by the BESIII and LHCb experiments.

In this paper, we construct and train end-to-end neural network-based dialogue systems usingan updated version of the recent Ubuntu Dialogue Corpus, a dataset containing almost 1 million multi-turn dialogues, with a total of over 7 million utterances and 100 million words. This dataset is interesting because of its size, long context lengths, and technical nature; thus, it can be used to train large models directly from data with minimal feature engineering, which can be both time consuming and expensive. We provide baselines  in two different environments: one where models are trained to maximize the log-likelihood of a generated utterance  conditioned on the context of the conversation, and one where models are trained to select the correct next response from a list of candidate responses. These are both evaluated on a recall task that we call Next Utterance Classification (NUC), as well as other generation-specific metrics. Finally, we provide a qualitative error analysis to help determine the most promising directions for future research on the Ubuntu  Dialogue Corpus, and for end-to-end dialogue systems in general.

We study the semileptonic decays B c → B n ℓ + ν ℓ where B c(n) is the anti-triplet-charmed (octet) baryon with the SU (3) f flavor symmetry and helicity formalism. In particular, we present the decay branching ratios of B c → B n ℓ + ν ℓ in three scenarios: (a) an exact SU (3) f symmetry with equal masses for the anti-triplet-charmed (octet) baryon states of B c (B n ), (b) SU (3) f parameters without the baryonic momentum-transfer dependence, and (c) SU (3) f with baryonic transition form factors in the heavy quark limit. We show that our results are all consistent with the existing data. Explicitly, we predict that B(Ξ + c → Ξ 0 e + ν e ) = (11.9 ± 1.3, 9.8 ± 1.1, 10.7 ± 0.9) × 10 −2 and B(Ξ 0 c → Ξ − e + ν e ) = (3.0 ± 0.3, 2.4 ± 0.3, 2.7 ± 0.2) × 10 −2 in the scenarios (a), (b) and (c) agree with the data of (14.0 +8.3 −8.7 ) × 10 −2 and (5.6 ± 2.6) × 10 −2 from the CLEO Collaboration, respectively. In addition, we obtain that B(Λ + c → ne + ν e ) = (2.8 ± 0.4, 4.9 ± 0.4, 5.1 ± 0.4) × 10 −3 in (a), (b) and (c).We also examine the longitudinal asymmetry parameters of α(B c → B n ℓ + ν ℓ ), which are sensitive to the different scenarios with SU (3) f . Some of the decay branching ratios and asymmetries can be observed by the ongoing experiments at BESIII and LHCb as well as the future searches by BELLEII.