LLM benchmark

BIG-Bench: A new benchmark for assessing LLMs, offering challenging, long-lasting tests to evaluate capabilities and biases comprehensively.

AGIEval is a human-centric benchmark that evaluates the general abilities of foundation models in tasks pertinent to human cognition and problem-solving.

ARC-c is a challenging variation of the ARC Benchmark, designed to assess the reasoning and commonsense understanding of large language models. Learn more about this dataset and the difficulty it presents for models.