🌌 FinErva: Interpretable Multimodal Reasoning for Robo Advisory A dataset & lightweight training framework that teaches small models to think like financial analysts. 🔗 Code Repository: GitHub – FinErva Framework FinErva — short for FINancial llm with minERVA wisdom — is a multimodal Chain of Thought (CoT) dataset designed explicitly for financial reasoning. It captures two of the most economically important tasks in investment decision making: Contract & disclosure understanding (FinErva Pact) Candlestick chart technical analysis (FinErva Price) And here’s the bigger reveal: FinErva enables models under 0.8B parameters to approach the reasoning ability of human finance professionals — including step by step interpretability — while remaining cost efficient and deployment friendly. This dataset aims to support research on auditable, multimodal, interpretable, and financially compliant AI systems for robo advisory, risk management, and financial decision support. 🎯 Key Features 🧠 Multimodal Chain of Thought (CoT) The first financial dataset combining contracts, real world financial images, and candlestick charts with human verified reasoning chains . 📊 Realistic Financial Conte…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy