arXiv subjects

Hongyuan Lu

Publications and source records attributed to Hongyuan Lu.

23 records · Page 2Linked to original sources

arXiv2023

Advancing Multilingual Pre-training: TRIP Triangular Document-level Pre-training for Multilingual Language Models

Despite the success of multilingual sequence-to-sequence pre-training, most existing approaches rely on document-level monolingual corpora in many different languages, sentence-level bilingual corpora,\footnote{In this paper, we use `bilingual corpora' to denote parallel corpora with `bilingual translation pairs' in many different language pairs, each consisting of two sentences/documents with the same meaning written in different languages. We use `trilingual corpora' to denote parallel corpora with `trilingual translation pairs' in many different language combinations, each consisting of three sentences/documents.} and sometimes synthetic document-level bilingual corpora. This hampers the performance with cross-lingual document-level tasks such as document-level translation. Therefore, we propose to mine and leverage document-level trilingual parallel corpora to improve sequence-to-sequence multilingual pre-training. We present \textbf{Tri}angular Document-level \textbf{P}re-training (\textbf{TRIP}), which is the first in the field to accelerate the conventional monolingual and bilingual objectives into a trilingual objective with a novel method called Grafting. Experiments show that TRIP achieves several strong state-of-the-art (SOTA) scores on three multilingual document-level machine translation benchmarks and one cross-lingual abstractive summarization benchmark, including consistent improvements by up to 3.11 d-BLEU points and 8.9 ROUGE-L points.

cs.CL↗

arXiv2023

PCC: Paraphrasing with Bottom-k Sampling and Cyclic Learning for Curriculum Data Augmentation

Curriculum Data Augmentation (CDA) improves neural models by presenting synthetic data with increasing difficulties from easy to hard. However, traditional CDA simply treats the ratio of word perturbation as the difficulty measure and goes through the curriculums only once. This paper presents \textbf{PCC}: \textbf{P}araphrasing with Bottom-k Sampling and \textbf{C}yclic Learning for \textbf{C}urriculum Data Augmentation, a novel CDA framework via paraphrasing, which exploits the textual paraphrase similarity as the curriculum difficulty measure. We propose a curriculum-aware paraphrase generation module composed of three units: a paraphrase candidate generator with bottom-k sampling, a filtering mechanism and a difficulty measure. We also propose a cyclic learning strategy that passes through the curriculums multiple times. The bottom-k sampling is proposed to generate super-hard instances for the later curriculums. Experimental results on few-shot text classification as well as dialogue generation indicate that PCC surpasses competitive baselines. Human evaluation and extensive case studies indicate that bottom-k sampling effectively generates super-hard instances, and PCC significantly improves the baseline dialogue agent.

cs.CL↗

arXiv2021

Partner Personas Generation for Diverse Dialogue Generation

Incorporating personas information allows diverse and engaging responses in dialogue response generation. Unfortunately, prior works have primarily focused on self personas and have overlooked the value of partner personas. Moreover, in practical applications, the availability of ground truth partner personas is often not the case. This paper attempts to tackle these issues by offering a novel framework that leverages automatic partner personas generation to enhance the succeeding dialogue generation. We incorporate reinforcement learning with a dedicatedly designed critic network for reward judgement. Experimental results from both automatic and human evaluation demonstrate a) Our framework is capable of generating relevant, informative and coherent partner personas, even compared to the ground truth partner personas. b) Generated partner personas enhance the succeeding response generation, thus surpassing our baselines and comparison model when partner personas are missing during the inference stage. c) Our framework generates responses that are more informative and engaging than our baseline conditioned on the ground truth partner personas during inference. d) Our dedicatedly designed critic network reinforces our framework effectively. Finally, our framework gives better explainability and reduces the demands for external databases for partner personas.

cs.CL↗

arXiv2018

Thermodynamics analysis on BaF2 intermediate phase in solution-derived YBCO superconducting film deposition

In the YBa2Cu3O7-δ (YBCO) high temperature superconducting thin film fabrication via the chemical solution deposition method, BaF2 is an important intermediate phase during heat treatment. In this paper, BaF2 thermodynamics stability was analyzed through calculating the standard Gibbs free energy change (ΔGT) of the reactions related to other intermediate phases within the temperature range of 700-1000 K. Two thermodynamics methods, the Gibbs free energy function method and standard formation molar Gibbs free energy method, were utilized to obtain the ΔGT values. The formation priority of BaF2 relative to other intermediate phases were verified at higher temperatures, while the possibility of BaCO3 formation was found at 700 K.

cond-mat.supr-con↗

arXiv2016

Critical Current Survival in YBCO Superconducting Layer of the Delaminated Coated Conductor

High temperature superconducting coated conductor (CC) could be practically applied in electric equipment due to its favorable mechanical properties and the critical current performance of YBCO superconducting layer. It is well known that CC could be easily delaminated because of its poor stress tolerance in thickness direction, i.e. along the c-axis of YBCO. Commonly, a stack including YBCO layer and silver stabilizer could be obtained after the delamination. It would be interesting to investigate the superconducting properties of the delaminated stack, since it could also be considered as a new type of CC with the silver stabilizer as the buffer layer, which is quite different from the oxide buffer layers in the traditional CC and might lead to new applications. In this study, a CC sample was delaminated by liquid nitrogen immersing. A Hall probe scanning system was employed to measure the critical current (IC) distribution of the original sample and the obtained stack. It was found that IC could be partially preserved after the delamination. Dense and crack-free morphologies of the delaminated surfaces were observed by scanning electron microscopy, and the potential application of the obtained stack in superconducting joint technology was discussed.

cond-mat.supr-con↗