ARC-C

ARC-C is a benchmark dataset that contains multiple-choice questions designed to evaluate the reasoning abilities of language models.

Papers using ARC-C (5)

HRM-Text: Efficient Pretraining Beyond Scaling2026On Robustness and Reliability of Benchmark-Based Evaluation of LLMs2025Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts2025Self-Reasoning Language Models: Unfold Hidden Reasoning Chains with Few Reasoning Catalyst2025TODO: Enhancing LLM Alignment with Ternary Preferences2024