FinDialogLens uses GPT-4o to extract trade events from financial chatrooms
FinDialogLens is a hybrid LLM pipeline designed to extract events from multi-party financial chatrooms, specifically to identify missed trades. It uses compact fine-tuned classifiers as scaffolds to detect RFQ-triggers and price/trade outcome metadata, segments RFQ windows per event, and fills argument roles via a Trade Engine. With GPT-4o, it achieves 92.1% accuracy on final price and 94.3% on…
1 source primary source
Key points
- FinDialogLens reaches 92.1% accuracy on final price and 94.3% on trade outcome using GPT-4o
- Fine-tuned open-source LLMs with as few as 3B parameters achieve comparable performance
- Difficulty-aware router cuts LLM calls by 85% on final price, saving over $300/day at 70,000 RFQ/day scale
Read the original at arXiv cs.CL · by Chin-Lun Fu, Hong Ni, Behrouz Madahian primary sourceOpen source ↗
Topics · follow one to build your own front page
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Enterprise & Industry
All →- Capcom’s RE Engine to integrate AI for game development workflows · 1 src
- A look at a radio show co-hosted by an AI DJ and a human that is expanding in LA and other cities, as radio workers worry synthetic hosts may soon replace them · 1 src
- Microsoft reports AI gives attackers early advantage in cyber threat landscape · 5 src
- Airbnb expands access to OpenAI’s GPT-6 Astra for bug fixes and system design · 1 src
- IBM releases self-hosted IBM Bob for on-prem and air-gapped use · 1 src
Comments
via GitHub Discussions