Google

A practical application of a large language model is to use a direct instruction, such as 'Extract the main financial figures from the following earnings report,' to parse a given text. For instance, when applied to a report stating 'The company reported a revenue of \$10 million in the first quarter with a profit margin of 15%...', the model would identify and return the key data points in a structured format: 'Revenue: \$10 million, Profit Margin: 15%'.

Example of Financial Data Extraction Using a Prompt

Based on the case study below, analyze why the initial instruction given to the language model was ineffective and write a revised, more precise instruction that would be more likely to produce only the desired output.

Refining a Data Extraction Prompt

You are an analyst tasked with using a language model to quickly summarize key metrics from corporate press releases. Read the text below and write a single, clear instruction (a prompt) that would direct a language model to extract *only* the quarterly revenue and the net profit figures.

Text: 'In its Q2 earnings call, TechSolutions Inc. reported a strong performance, with quarterly revenue reaching an impressive $45.2 million, a 12% increase from the previous year. The company's net profit for the quarter was $8.5 million. Additionally, operating expenses were kept at $20 million, and the company announced a new partnership with a major cloud provider.'

Crafting a Financial Data Extraction Prompt

A language model is given the instruction: 'From the following text, extract the Total Revenue, Net Income, and Earnings Per Share (EPS) into a structured list.' Which of the following outputs best represents the model's successful execution of this instruction on the text below?

Text: 'For the third quarter, Innovate Corp. announced a significant 20% year-over-year growth, achieving a total revenue of $500 million. After accounting for operational costs of $350 million and taxes, the company's

Learn Before

Related