EricLu created a dataset of scientific problems paired with solutions. A newer version, SCP-378K, contains 377,705 examples and provides extracted source solutions for every problem. The dataset was last updated on May 8, 2026.
Use Cases
- Train models for scientific question answering based on problem-solution pairs.
- Benchmark solution generation algorithms against a standard solution corpus.
- Analyze the structure of scientific problems and their corresponding solutions.
Strengths
- Dataset is part of a series with a newer, larger version (SCP-378K) containing 377,705 examples.
- Author is identified (EricLu), providing a point of contact.
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
- EricLu
- Freshness
- Last updated 2026-05-08 06:04:15; freshness should be verified.