Learning To Replicate Expert Judgment In Financial Tasks
Source
Thinking Machines editorial sitemap
Author
Thinking Machines editorial sitemap
Date
Terms in this piece · Glossary
distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
Why it matters
A concrete recipe for encoding expert judgment into a small model: high-quality human annotations plus asymmetric-clipping CISPO and on-policy distillationTraining a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.Full definition →, outperforming frontier models on accuracy and recall at a fraction of inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → cost.
articlePutting Task Expertise Into RlThinking Machines editorial sitemap
articleAre the Financial Reasoning from LLMs Credible? A Real World Test over Long-Horizon StatementsXinke Tong, Xuanming Zhang, Tianyi Tang, An Yang, Jiatu Hu, Guojie Lin, Zhenzhen Shi, Lingfeng Zeng, Boyu Yang, Bing Zhao, Hu Wei, Lin Qu, Dayiheng Liu
articleConfidence Estimation for Financial Vision-Language Models in Chart and Document UnderstandingReza Khanmohammadi, Simerjot Kaur, Charese H. Smiley, Ivan Brugere, Mohammad M. Ghassemi