paper-figure

Generate publication-quality plots by reading experiment logs and data files to produce reproducible plotting scripts and PDF outputs. This workflow applies consistent fonts, DPI, and publication styling to create print-ready figures for ML papers.

Create LaTeX tables from CSV or JSON experiment data by converting comparison metrics and ablation results into LaTeX-ready table code. This generates include snippets for downstream paper writing without manual formatting.

Yes, you can preserve manually created architecture figures. The workflow uses a PAPER_PLAN.md figure plan to distinguish between data-driven plots to generate automatically and manual figures to leave untouched.

Automate reproducible figure generation by using a PAPER_PLAN.md file to define the figure plan and reading JSON, CSV, or log data files. The workflow outputs reproducible plotting scripts alongside PDF vector outputs to ensure consistency.

Yes, a PAPER_PLAN.md figure plan is required to guide the generation process. It determines which data plots and comparison tables to produce automatically and which figures require manual creation.

Standard ML paper plots including line charts, scatter plots, heatmaps, and box or violin plots can be generated from experiment results. The workflow applies publication-oriented visual style with PDF vector output as the preferred format.