TechcraftingAI NLP

TechcraftingAI NLP

By Brad EdwardsTechnology
Download on the App Store

All TechcraftingAI NLP episodes

  • Ep. 89 - December 22, 2023

    arXiv NLP research summaries for December 22, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


    Today's research themes (AI summary)

    • Improving simultaneous machine translation using language models for branch prediction.
    • Linking insufficient equivariance in language models to hallucinations.
    • Automating data retrieval for English-Hindi cross-lingual summarization.
    • Enhancing Chinese conversational capabilities of LLMs via instruction tuning.
    • Identifying speakers in literature via prompt-based generation.
    • 20 min
    • Ep. 88 - December 21, 2023

      arXiv NLP research summaries for December 21, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


      Today's research themes (AI summary)

      • Large language models for speech translation and counterargument generation
      • Evaluating dialogue systems with metrics aligned to real-world usage
      • Injecting knowledge into language models for biomedical NLP
      • Analyzing model confidence and calibration in low-resource scenarios
      • Assessing emphasis transfer in speech-to-speech models
      • 21 min
      • Ep. 87 - December 20, 2023

        arXiv NLP research summaries for December 20, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


        Today's research themes (AI summary)

        • Evaluating medical large language models on Chinese benchmark datasets constructed from exams and clinical cases
        • Improving consistency and user satisfaction in dialogue systems with scenarios and multimodal feedback
        • Distilling complex reasoning from large language models using both positive and negative data
        • Analyzing the linguistic origins of English place names using statistical comparisons to other languages
        • Developing resources and models for Dutch language large language modeling
        • 26 min
        • Ep. 86 - December 19, 2023

          arXiv NLP research summaries for December 19, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


          Today's research themes (AI summary)

          • Fake news detection with augmented fact-checking by AI models
          • Punctuation and spacing prediction for historical Korean documents
          • Efficient inference methods for accelerating language models
          • Knowledge tracing focused on question difficulty prediction
          • Polyphone disambiguation for Chinese text-to-speech using external knowledge
          • 32 min
          • Ep. 85 - December 18, 2023

            arXiv NLP research summaries for December 18, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


            Today's research themes (AI summary)

            • Language model improvements, including multimodality, reasoning, and scalability
            • Bias and fairness considerations for language models, especially regarding gender
            • Retrieval augmented language models for improved grounding
            • Applying language models to specialized domains like healthcare and science
            • Model interpretability and analysis of language model behavior

            • 1 hr 10 min
            • Ep. 84 - December 17, 2023

              arXiv NLP research summaries for December 17, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


              Today's research themes (AI summary)

              • Chart question answering and summarization using few-shot prompting of large language models
              • Detecting AI-generated text by analyzing language traits
              • Extracting hyperparameters from scientific publications using models and datasets
              • Privacy-preserving federated learning for multilingual knowledge graph completion
              • Enhancing BERT for health applications via knowledge and model fusion
              • 7 min
              • Ep. 83 - December 16, 2023

                arXiv NLP research summaries for December 16, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                Today's research themes (AI summary)

                • Prompt engineering for few-shot learning and controllable generation
                • Multimodal modeling for sarcasm detection and dialogue systems
                • Graph neural networks for relation modeling in dialogues
                • Contrastive learning for representation debiasing and regularization
                • Domain evaluation of language models beyond perplexity
                • 17 min
                • Ep. 82 - December 15, 2023

                  arXiv NLP research summaries for December 15, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                  Today's research themes (AI summary)

                  • Improving language models with additional linguistic information and structure
                  • Using language models for specialized domains like medicine and accessibility
                  • Analyzing biases and inconsistencies in multilingual language models
                  • Employing language models for conversation and dialog tasks
                  • Enhancing question answering with external knowledge and multi-step reasoning
                  • 21 min
                  • Ep. 81 - December 14, 2023

                    arXiv NLP research summaries for December 14, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                    Today's research themes (AI summary)

                    • Evaluating language model susceptibility to misinformation in persuasive dialogue
                    • Mitigating dataset bias in language models for robust natural language understanding
                    • Using language models for social science measurement tasks like ideological scaling
                    • Enhancing reasoning capabilities of language models on semi-structured tabular data
                    • Developing verifiable text generation models to reduce hallucination
                    • 22 min
                    • Ep. 80 - December 13, 2023

                      arXiv NLP research summaries for December 13, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                      Today's research themes (AI summary)

                      • Text watermarking for detecting generative model outputs
                      • Robust few-shot named entity recognition
                      • Legal judgment prediction for Indian legal documents
                      • Prompting large language models for improved performance
                      • Relation extraction from biomedical text
                      • 18 min