- Is GPT-3 a Good Data Annotator? [paper]
- ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks [paper]
- Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction [paper]
- ZCODE4STRUCT: Code Generation for Few-Shot Structured Prediction from Natural Language [paper]
- Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction [paper]
- Zero-Shot Information Extraction via Chatting with ChatGPT [paper][code]
- Large Language Model Is Not a Good Few-shot Information Extractor, but a Good Reranker for Hard Samples! [paper]
- Thinking about GPT-3 in-context learning for biomedical IE? [paper[ code
- Lauge language models are fewshot clinical Information Extractors [paper]
- Zero-shot Clinical Entity Recognition using ChatGPT [paper]
- Yes but.. Can ChatGPT Identify Entities in Historical Documents? [paper]
- Evaluation-of-ChatGPT-on-Information-Extraction [paper]
- Improving Chatbot Responses with Information Extraction [paper]
- https://huggingface.co/spaces/Shad0ws/Information_Extraction_with_ChatGPT
- GPT4IE [code]
- ChatGPT as a Factual Inconsistency Evaluator for Abstractive Text Summarization [paper]
- Large Language Models are Diverse Role-Players for Summarization Evaluation [paper]
- Towards Making the Most of ChatGPT for Machine Translation [paper]
- Error Analysis Prompting Enables Human-Like Translation Evaluation in Large Language Models: A Case Study on ChatGPT [paper]