ArXiv NLP research for Thursday, June 13, 2024.

00:20: Chain-of-Though (CoT) prompting strategies for medical error detection and correction

01:31: CoastTerm: a Corpus for Multidisciplinary Term Extraction in Coastal Scientific Literature

02:52: RH-SQL: Refined Schema and Hardness Prompt for Text-to-SQL

04:01: Chain of Preference Optimization: Improving Chain-of-Thought Reasoning in LLMs

05:24: Leveraging Explicit Reasoning for Inference Integration in Commonsense-Augmented Dialogue Models

06:38: Investigating the translation capabilities of Large Language Models trained on parallel data only

07:56: LASER: Learning by Aligning Self-supervised Representations of Speech for Improving Content-related Tasks

09:09: DefAn: Definitive Answer Dataset for LLMs Hallucination Evaluation

11:20: Test of Time: A Benchmark for Evaluating LLMs on Temporal Reasoning

12:46: Orthogonality and isotropy of speaker and phonetic information in self-supervised speech representations

13:53: Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn't

14:47: ReadCtrl: Personalizing text generation with readability-controlled instruction learning

16:32: Self-Training for Sample-Efficient Active Learning for Text Classification with Pre-Trained Language Models

17:49: Sharing Matters: Analysing Neurons Across Languages and Tasks in LLMs

19:18: End-to-end Streaming model for Low-Latency Speech Anonymization

20:22: Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback

22:25: On the Effects of Heterogeneous Data Sources on Speech-to-Text Foundation Models

23:33: Understanding Jailbreak Success: A Study of Latent Space Dynamics in Large Language Models

24:35: Exploring Spoken Language Identification Strategies for Automatic Transcription of Multilingual Broadcast and Institutional Speech

25:47: AlignMMBench: Evaluating Chinese Multimodal Alignment in Large Vision-Language Models

27:15: Transformers meet Neural Algorithmic Reasoners

28:32: REVS: Unlearning Sensitive Information in Language Models via Rank Editing in the Vocabulary Space

30:02: Learning from Natural Language Explanations for Generalizable Entity Matching

31:14: ProxyLM: Predicting Language Model Performance on Multilingual Tasks via Proxy Models

32:29: DiscreteSLU: A Large Language Model with Self-Supervised Discrete Speech Units for Spoken Language Understanding

33:43: Improving Autoregressive Training with Dynamic Oracles