The paper explores the potential of using the Llama 2 Large Language Model (LLM) for multitask analysis of financial news. The authors employed a fine-tuning approach based on PEFT/LoRA. During the study, the model was fine-tuned for several tasks, including analyzing financial texts, highlighting key points, summarizing texts, and extracting named entities with their corresponding sentiments. The results demonstrated that the fine-tuned Llama 2 model can effectively carry out multitask financial news analysis, generating structured text as well as JSON-formatted data for further processing. The extracted sentiments for named entities can also be utilized as predictive features in supervised machine learning models with quantitative target variables.
Leveraging Fine-Tuned Llama 2 GPT Model for Analyzing Financial News (arXiv:2308.13032v1 [cs.CL])
