zjunlp/OceanBenchmark
收藏资源简介:
OceanBenchmark是一个基准数据集,旨在评估以海洋为重点的大型模型的综合能力。它涵盖了从单模态海洋科学知识问答到复杂的多模态视觉问答的多种任务。数据集包括多个子集(Science-Text、Science-MM、Sonar、Bio),每个子集有不同的任务类型(QA、VQA)和样本大小。每个子集都有特定的特征和格式描述。数据集支持中文和英文,并采用MIT许可证。
OceanBenchmark is a benchmark dataset designed to assess the comprehensive capabilities of ocean-focused large models. It covers a wide range of tasks, spanning from unimodal ocean science knowledge question answering to complex multimodal visual question answering. The dataset consists of multiple subsets (Science-Text, Science-MM, Sonar, Bio), each with distinct task types (QA, VQA) and sample sizes. Each subset has its own specific characteristics and format specifications. The dataset supports both Chinese and English, and is distributed under the MIT License.




