Academicmcqa: MMLU Auxiliary Train Set for Reinforcement Learning

Academicmcqa is a reformatted version of the MMLU (Massive Multitask Language Understanding) auxiliary train set. The dataset was created by NousResearch for use in Atropos, their reinforcement learning environments framework. It was last updated on Hugging Face on April 28, 2025.

Use Cases

• Training reinforcement learning agents based on multiple-choice question answering.
• Benchmarking language model performance on academic knowledge tasks.
• Fine-tuning models for specific domains covered by the MMLU auxiliary set.

Strengths

• Dataset is derived from the established MMLU benchmark, suggesting a known quality standard.
• Explicitly formatted for compatibility with the Atropos RL framework, indicating a specific use case.

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

NousResearch

Collection Method

Reformatted from the MMLU (Massive Multitask Language Understanding) auxiliary train set.

Freshness

Last updated 2025-04-28 19:27:07; freshness should be verified.