gwiezdny/MTBench_finance_news

MTBench: A Multimodal Time Series Benchmark MTBench (Huggingface, Github, Arxiv) is a suite of multimodal datasets for evaluating large language models (LLMs) in temporal and cross-modal reasoning tasks across finance and weather domains. Each benchmark instance aligns high-resolution time series (e.g., stock prices, weather data) with textual context (e.g., news articles, QA prompts), enabling research into temporally grounded and multimodal understanding. 🏦 Labled… See the full description on the dataset page: https://huggingface.co/datasets/gwiezdny/MTBench_finance_news.