Bloomberg announced its own large language model trained on financial data to support NLP tasks within the industry.
The model, BloombergGPT,
is specific to the financial domain, making it able to analyze related
data and help make investment decisions and other tasks.
- Bloomberg
customers can use it to analyze data available on the Bloomberg
Terminal, its software system for financial market data and trades.
- Internally,
the financial data vendor could use the generative AI model in areas
like sentiment analysis, Q&A, and named entity recognition.
- The
50-billion parameter causal language model was trained on part of a
large training corpus with more than 700 billion tokens, compiled by
Bloomberg's ML product and research and AI engineering teams.
- Bloomberg
CTO Shawn Edwards said it's the "first LLM focused on the financial
domain," offering a "much higher performance out-of-the-box than custom
models for each application, at a faster time-to-market."
