FutureX-Past: Past Questions from a Live LLM Future Prediction Benchmark

Futurex-Past is a dataset of past questions from the FutureX benchmark, which evaluates the future prediction capabilities of Large Language Model agents. The dataset is hosted by futurex-ai on Hugging Face and was last updated on May 10, 2026. FutureX features an automated pipeline that generates questions about upcoming real-world events.

Use Cases

• Benchmarking LLM agent performance on future event prediction based on the described benchmark structure
• Analyzing trends in question formulation for automated future prediction evaluations
• Training models to predict outcomes of real-world events based on historical benchmark questions

Strengths

• Dataset is part of a live, dynamic benchmark with an automated scoring pipeline, as described.
• The repository was last updated on 2026-05-10, suggesting ongoing maintenance.

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

futurex-ai

Collection Method

Generated via an automated pipeline for the FutureX benchmark.

Freshness

Last updated 2026-05-10 10:57:45; freshness should be verified.