BrainBench

该数据集包含了评估人类参与者与语言模型在决策任务中协作效果的性能指标和LLM困惑度得分。此外，该数据集在研究中被用于分析人类与机器判断的互补性，重点关注信心校准和分类多样性。该数据集的任务是评估在决策任务中，人类与语言模型有效协作的条件。This dataset includes performance metrics for assessing the collaborative effectiveness between human participants and large language models (LLMs) in decision-making tasks, as well as LLM perplexity scores. Additionally, this dataset was utilized in the study to analyze the complementarity between human and machine judgments, with a focus on confidence calibration and classification diversity. The task of this dataset is to evaluate the conditions under which humans and language models can collaborate effectively in decision-making tasks.

搜集汇总

背景与挑战

背景概述

BrainBench是一个用于评估大语言模型在预测神经科学实验结果方面性能的基准测试数据集。它包含完整的代码库，支持实验分析、模型微调和结果可视化，旨在比较LLM与人类专家在科学发现任务中的表现。该数据集基于已发表的研究论文，专注于推动神经科学领域的预测性研究。