TechcraftingAI NLP

TechcraftingAI NLP

By Brad EdwardsTechnology
Download on the App Store

All TechcraftingAI NLP episodes

  • Ep. 99 - January 1, 2024

    arXiv NLP research summaries for January 1, 2024. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


    Today's research themes (AI summary)

    • Sentiment analysis of religious texts using pre-trained language models
    • Evaluating controllable text generation by large language models under diverse instructions
    • Assessing large language models on complex named entity recognition with model engineering techniques
    • Evaluating tool learning capabilities of large language models in real-world scenarios
    • Parameter-efficient instruction tuning methods for large language models on code tasks
    • 17 min
    • Ep. 98 - December 31, 2023

      arXiv NLP research summaries for December 31, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


      Today's research themes (AI summary)

      • Multi-modal intent detection using shallow-to-deep interaction frameworks and data augmentation.
      • Knowledge-based question answering by decomposing questions, retrieving entities, and reasoning over knowledge graphs.
      • Specializing large language models for domain-specific tasks like geoscience through further pretraining and instruction tuning.
      • Evaluating language models more robustly using diverse prompts and tailored metrics for specific use cases.
      • Incorporating personality, persona, and profile modeling into conversational agents and chatbots.
      • 18 min
      • Ep. 97 - December 30, 2023

        arXiv NLP research summaries for December 30, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


        Today's research themes (AI summary)

        • Reasoning and inference in language models for knowledge graph QA
        • Mitigating false negatives in dense retrieval for QA
        • Named entity recognition for social media and low-resource languages
        • Evaluating coreference resolution in literary texts
        • Improving language models for speech synthesis
        • 20 min
        • Ep. 96 - December 29, 2023

          arXiv NLP research summaries for December 29, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


          Today's research themes (AI summary)

          • Improving language model performance on specialized tasks like medical language understanding and quantitative reasoning.
          • Enhancing language model capabilities for collaboration, teamwork, and dialog.
          • Developing more explainable and interpretable language models.
          • Applying language models to information extraction tasks.
          • Evaluating multimodal language models on reasoning and commonsense tasks.
          • 20 min
          • Ep. 95 - December 28, 2023

            arXiv NLP research summaries for December 28, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


            Today's research themes (AI summary)

            • Improving language models through new pretraining objectives and techniques like bidirectional alignment.
            • Evaluating capabilities of large language models through new benchmarks focused on reasoning, consistency, and robustness.
            • Applying large language models to new domains like scientific computing and causal analysis.
            • Controlling text generation by modelling stylistic features and mitigating hallucination.
            • Developing metrics to assess qualities like coherence in generated text.
            • 24 min
            • Ep. 94 - December 27, 2023

              arXiv NLP research summaries for December 27, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


              Today's research themes (AI summary)

              • Knowledge transfer for generalized category discovery and novel categories
              • Graph-based methods for code clone detection
              • Understanding news creation intent
              • Converting single-turn to multi-turn data for conversational QA
              • Metrics for conceptual diversity in text
              • 18 min
              • Ep. 93 - December 26, 2023

                arXiv NLP research summaries for December 26, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                Today's research themes (AI summary)

                • Improving language models through techniques like contrastive learning, knowledge distillation, and compositional training.
                • Leveraging knowledge graphs and retrieval to enhance reasoning capabilities of language models.
                • Developing language models for new languages and domains like Uzbek and computer security.
                • Mitigating issues with language models like hallucination and backdoor attacks.
                • Applying language models to tasks like medical report generation, temporal knowledge graphs, and scientific information extraction.
                • 18 min
                • Ep. 92 - December 25, 2023

                  arXiv NLP research summaries for December 25, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                  Today's research themes (AI summary)

                  • Improving language model performance on specific domains like e-commerce and computer security.
                  • Enhancing language model capabilities for knowledge reasoning and factual accuracy.
                  • Applying language models to tasks like medical report generation and machine translation.
                  • Developing techniques to align large language models to specific end tasks.
                  • Leveraging multi-modal inputs like images and video for language tasks like sign language translation.
                  • 25 min
                  • Ep. 91 - December 24, 2023

                    arXiv NLP research summaries for December 24, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                    Today's research themes (AI summary)

                    • Evaluating and mitigating biases in large language models from a group fairness perspective.
                    • Adapting large language models for dense retrieval through additional pre-training.
                    • Developing universal information extraction models for both English and Chinese.
                    • Using multi-granularity embeddings and enhanced labeling for biomedical named entity recognition.
                    • Generating simplified lay definitions of medical jargon for patient education.
                    • 11 min
                    • Ep. 90 - December 23, 2023

                      arXiv NLP research summaries for December 23, 2023. You can find summaries and links to each article ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.


                      Today's research themes (AI summary)

                      • Improving language model performance via techniques like depth up-scaling and self-supervised pre-training.
                      • Detecting and mitigating bias in language models for Indian languages and gender fairness.
                      • Using large language models for tasks like question answering, dialogue, and evaluating dialog systems.
                      • Representing emotions and sentiment in speech with pre-trained models.
                      • Inductive learning of context free grammars from textual data.
                      • 16 min