jablonkagroup/corral-oss-trace-logprobs

该数据集包含多个子数据集，涵盖催化剂、机器学习、电阻器、逆合成和光谱等不同领域的任务和子任务。每个子数据集包含训练集，记录了任务、试验、分数、环境、级别、代理类型、消息编号、试验步骤、消息ID等多个特征字段。数据集主要用于记录和分析不同任务和试验中的性能指标和日志信息。

This dataset contains multiple sub-datasets covering tasks and subtasks in various domains such as catalyst, machine learning, resistor, retrosynthesis, and spectra. Each sub-dataset includes a training set with features like task, trial, score, environment, level, agent type, message number, step in trial, message ID, etc. The dataset is primarily used to record and analyze performance metrics and log information across different tasks and trials.