Carbon Accounting Management Platform Benchmark…
CCR Re -LLM for augmented actuarial science
Empowering Actuarial Precision: Harnessing Advanced LLMs with AWS for Dynamic Risk Assessment at CCR Re
Business requirements
CCR Re wished to leverage innovative tools to quickly measure its risk exposure to sudden events and to adapt to global events such as riots, war risks and pandemics. Together with AWS, we have fine-tuned large language models (LLMs) to act as helpful actuarial assistants.
Technical implementation
Starting from GPT-like models trained on a mix of publicly available, synthetic datasets such as HuggingFace’s Zephyr series, we relied extensively on fully managed services such as Amazon SageMaker for subsequent fine-tuning on proprietary data to build then serve a high-performance yet cost-effective model.
