eltociear/matplotlib-tasks-v1

matplotlib-tasks-v1 Task dataset for a Matplotlib/Seaborn RL / eval environment, in the shape used by the Prime Intellect Environments Hub. 27 plotting tasks over a fixed set of small datasets. The model builds the figure, then reads the answer back off the axes; the grade is an exact row comparison against a reference. Deterministic — no LLM judge, no external API, no network, and no pixels. Category Tasks Covers line_plots 6 ydata/xdata readback, multiple lines… See the full description on the dataset page: https://huggingface.co/datasets/eltociear/matplotlib-tasks-v1.