amazon-agi/Amazon-Nova-1.0-Lite-evals

该数据集包含了用于评估Amazon Nova Lite模型的多个公共基准数据集，包括MMLU、DROP、GPQA、MATH、GSM8K、IFEval、BBH、HumanEval、FinQA、FLORES、MMMU、ChartQA、DocVQA、TextVQA、VATEX、EgoSchema、VisualWebBench、MM-Mind2Web和GroundUI-1K。每个数据集都有特定的评估方法，如0-shot或6-shot CoT（Chain-of-Thought）提示，并报告了不同的评估指标，如准确率、F1分数、CIDEr分数等。

This dataset includes multiple public benchmark datasets used to evaluate the Amazon Nova Lite model, such as MMLU, DROP, GPQA, MATH, GSM8K, IFEval, BBH, HumanEval, FinQA, FLORES, MMMU, ChartQA, DocVQA, TextVQA, VATEX, EgoSchema, VisualWebBench, MM-Mind2Web, and GroundUI-1K. Each dataset has specific evaluation methods, such as 0-shot or 6-shot CoT (Chain-of-Thought) prompting, and reports different evaluation metrics, such as accuracy, F1 score, CIDEr score, etc.