MTBench Finance Aligned Pairs Long: Stock Prices Aligned with Textual Context

MTBench is a multimodal time series benchmark for evaluating large language models in temporal and cross-modal reasoning. The dataset aligns high-resolution financial time series, such as stock prices, with textual context like news articles or QA prompts. It was created by GGLabYale and last updated on 2025-05-23.

Use Cases

• Benchmarking LLM performance on temporal reasoning tasks based on aligned stock price and news data.
• Training models for cross-modal understanding based on the alignment of time series data with textual prompts.
• Researching temporally grounded language models based on the described multimodal finance instances.

Strengths

• Designed specifically for evaluating LLMs on multimodal and temporal reasoning tasks.
• Aligns high-resolution time series data with textual context, enabling a specific research focus.
• Part of a published benchmark suite with associated documentation on Huggingface, Github, and Arxiv.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

GGLabYale

Freshness

Last updated 2025-05-23 07:54:36; freshness should be verified.