FInstruct contains instruction-following data extracted from recent research articles by MSCI, S&P, and arXiv. The data set was generated in the style of self-instruct using the model behind HuggingChat. I'm actively curating this data set by adding quality labels and improving the generation process.
The quality label:
- 0: Not curated
- 1: Should be excluded
- 2: Pass basic cleansing
This data set may be a resource for researchers, academics, and industry professionals interested in exploring the latest research in quantitative finance. Additionally, it can be a training data set for those looking to fine-tune language models for financial analysis and decision-making.
If you notice any errors or have any suggestions, please open an issue or shoot me a message directly.