Skip to content

capwitf/My-MathModeling-skills / math-figure

math-figure

Unverified

Use when 高教社杯/CUMCM 数学建模论文需要规划、选择、审查或修复图、图表、表格转图、图注、图像证据行、正文放置、附录降级或图后结论。

FileKindSize
SKILL.mdskill md4.6 KB
agents/openai.yamlconfig312 B
assets/chart-templates/paper_style.pyscript8.9 KB
assets/high-impact-templates/chart-atlas/atlas-01-bar-charts.pngbinary122.4 KB
assets/high-impact-templates/chart-atlas/atlas-02-line-trends.pngbinary413.7 KB
assets/high-impact-templates/chart-atlas/atlas-03-heatmaps.pngbinary156.1 KB
assets/high-impact-templates/chart-atlas/atlas-04-scatter-bubble.pngbinary473.7 KB
assets/high-impact-templates/chart-atlas/atlas-05-radar-polar.pngbinary615.3 KB
assets/high-impact-templates/chart-atlas/atlas-06-distributions.pngbinary228.3 KB
assets/high-impact-templates/chart-atlas/atlas-07-forest-interval.pngbinary263.8 KB
assets/high-impact-templates/chart-atlas/atlas-08-area-stacked.pngbinary326.9 KB
assets/high-impact-templates/chart-atlas/atlas-09-image-plates.pngbinary350.6 KB
assets/high-impact-templates/chart-atlas/atlas-10-network-matrix.pngbinary304.8 KB
assets/high-impact-templates/figures4papers/assets/Dispersion_motivation.pngbinary818.0 KB
assets/high-impact-templates/figures4papers/assets/Dispersion_observation.pngbinary790.0 KB
assets/high-impact-templates/figures4papers/assets/Dispersion_observation_distillation.pngbinary602.2 KB
assets/high-impact-templates/figures4papers/assets/ImmunoStruct_contrastive.pngbinary490.4 KB
assets/high-impact-templates/figures4papers/assets/ImmunoStruct_results_CEDAR.pngbinary1.2 MB
assets/high-impact-templates/figures4papers/assets/ImmunoStruct_results_IEDB.pngbinary933.9 KB
assets/high-impact-templates/figures4papers/assets/ImmunoStruct_schematic.pngbinary950.1 KB
assets/high-impact-templates/figures4papers/assets/RNAGenScape_schematic.pngbinary976.9 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/figures/ablation.pngbinary347.0 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/figures/comparison.pngbinary794.0 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/figures/comparison_human.pngbinary860.2 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/figures/comparison_worm.pngbinary903.6 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/plot_ablation.pyscript2.9 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/plot_comparison.pyscript3.8 KB
assets/high-impact-templates/figures4papers/figure_CellSpliceNet/plot_comparison_cross_species.pyscript6.4 KB
assets/high-impact-templates/figures4papers/figure_Cflows/diffusion_swiss_roll.pyscript3.5 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/diffusion_swiss_roll.pngbinary1.7 MB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/fig2_comparison_GeneRegulatory.pdfbinary24.2 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/fig2_comparison_GeneRegulatory.pngbinary280.7 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/fig2_comparison_Trajectory.pdfbinary19.3 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/fig2_comparison_Trajectory.pngbinary379.2 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/figX_comparison_Ablation.pdfbinary18.8 KB
assets/high-impact-templates/figures4papers/figure_Cflows/figures/figX_comparison_Ablation.pngbinary321.2 KB
assets/high-impact-templates/figures4papers/figure_Cflows/plot_comparison_Ablation.pyscript2.4 KB
assets/high-impact-templates/figures4papers/figure_Cflows/plot_comparison_GeneRegulatory.pyscript3.0 KB
assets/high-impact-templates/figures4papers/figure_Cflows/plot_comparison_Trajectory.pyscript2.9 KB
assets/high-impact-templates/figures4papers/figure_Dispersion/figures/idea.pngbinary375.8 KB
assets/high-impact-templates/figures4papers/figure_Dispersion/figures/illustration.pngbinary745.1 KB
assets/high-impact-templates/figures4papers/figure_Dispersion/plot_idea.pyscript2.4 KB
assets/high-impact-templates/figures4papers/figure_Dispersion/plot_illustration.pyscript15.1 KB
assets/high-impact-templates/figures4papers/figure_FPGM/figures/freq_prior.pngbinary626.0 KB
assets/high-impact-templates/figures4papers/figure_FPGM/plot_freq_prior.pyscript6.4 KB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/figures/bars_ablation_Cancer.pngbinary797.6 KB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/figures/bars_ablation_IEDB.pngbinary1.1 MB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/figures/bars_comparison_Cancer.pngbinary781.0 KB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/figures/bars_comparison_IEDB.pngbinary739.8 KB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/plot_bars.pyscript7.5 KB
assets/high-impact-templates/figures4papers/figure_ImmunoStruct/raw_data.pyscript4.4 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/figures/manifold.pngbinary105.7 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/figures/manifold_holes.pngbinary996.0 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/figures/results_comparison_optimization.pngbinary830.5 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/figures/results_comparison_speed.pngbinary90.2 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/figures/results_sweep.pngbinary138.8 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/plot_comparison.pyscript11.0 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/plot_hole_manifold.pyscript2.5 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/plot_manifold.pyscript1.8 KB
assets/high-impact-templates/figures4papers/figure_RNAGenScape/plot_sweep.pyscript2.8 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/figures/ablation_curves.pngbinary586.6 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/figures/comparison_posttraining.pngbinary243.4 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/figures/comparison_radar.pngbinary790.9 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/figures/concept.pngbinary1.6 MB
assets/high-impact-templates/figures4papers/figure_VIGIL/plot_ablation.pyscript5.9 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/plot_comparison_radar.pyscript7.2 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/plot_concept.pyscript9.0 KB
assets/high-impact-templates/figures4papers/figure_VIGIL/plot_posttraining.pyscript2.9 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/figures/brute_force.pngbinary1.7 MB
assets/high-impact-templates/figures4papers/figure_brainteaser/figures/correctness_by_category.pngbinary640.1 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/figures/correctness_by_subcategory.pngbinary1.0 MB
assets/high-impact-templates/figures4papers/figure_brainteaser/figures/rewriting.pngbinary400.5 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/figures/selfcorrection_math.pngbinary597.2 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/plot_brute_force.pyscript9.9 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/plot_correctness_by_category.pyscript6.3 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/plot_correctness_by_subcategory.pyscript6.4 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/plot_rewriting.pyscript3.6 KB
assets/high-impact-templates/figures4papers/figure_brainteaser/plot_selfcorrection_math.pyscript3.9 KB
assets/high-impact-templates/figures4papers/figure_ophthal_review/figures/composition_heatmap.pngbinary273.0 KB
assets/high-impact-templates/figures4papers/figure_ophthal_review/figures/trend_by_month.pngbinary355.6 KB
assets/high-impact-templates/figures4papers/figure_ophthal_review/plot_composition.pyscript3.0 KB
assets/high-impact-templates/figures4papers/figure_ophthal_review/plot_trend.pyscript5.4 KB
references/chart-template-index.mdmarkdown4.5 KB
references/contest-chart-selection.mdmarkdown2.5 KB
references/contest-figure-pitfalls.mdmarkdown2.0 KB
references/high-impact-chart-assets.mdmarkdown4.2 KB
references/paper-figure-style.mdmarkdown2.4 KB
references/selection-guide.mdmarkdown3.0 KB
references/visual-readiness-loop.mdmarkdown1.7 KB
scripts/check_contest_figure.pyscript3.0 KB
scripts/profile_result_table.pyscript3.8 KB

assets/high-impact-templates/figures4papers/figure_VIGIL/plot_ablation.py

View on GitHub
import os
import numpy as np
from matplotlib import pyplot as plt
from matplotlib import gridspec as gridspec


data_ablation = {
    'methods': [
        'DPO',
        'DA-DPO',
        'VIGIL (Ours)',
    ],
    'colors': [
        "#D88F8A",
        "#8BCF8B",
        "#0F4D92"
    ],
    '$\beta$': [0.05, 0.1, 0.2, 0.5],
    '$\lambda$': [0.1, 0.5, 1.0, 2.0],
    'data_fraction': [0.1, 0.25, 0.5, 1.0],
    'results': {
        'SFT': np.array([82.1]),  # POPE_Adv
        # This is 3 methods on 4 beta values.
        '$\beta$': np.array([
            [79.5, 82.8, 81.0, 76.2],
            [83.2, 84.2, 83.5, 81.0],
            [86.8, 86.9, 86.5, 85.8],
        ]),
        # This is our method.
        '$\lambda$': np.array([
            [84.5, 86.1, 86.9, 87.2], # POPE_Adv
            [49.8, 49.6, 49.5, 48.8], # MathVista
        ]),
        'data_fraction': np.array([
            [82.78, 83.19, 85.05, 85.67],
            [84.04, 85.46, 87.07, 87.81],
            [86.69, 88.18, 89.17, 89.82],
        ])
    }
}


def plot_curves(data_ablation):
    """
    Left panel: ablation on hyperparameter beta (x = beta values, y = score, 3 curves for 3 methods).
    Right panel: blank.
    """
    methods = data_ablation['methods']
    colors = data_ablation['colors']

    fig, axes = plt.subplots(1, 3, figsize=(27, 6), gridspec_kw={'width_ratios': [1.1, 1, 1]})

    ax = axes[0]
    y_ticks = [78, 82, 86, 90]
    data_fraction_key = 'data_fraction'
    data_fraction_vals = np.asarray(data_ablation[data_fraction_key])
    results_data_fraction = data_ablation['results'][data_fraction_key]  # shape (n_methods, n_data_fraction)
    x_pos = np.arange(len(data_fraction_vals))
    y_sft = data_ablation['results']['SFT']
    # SFT only horizontal line.
    ax.plot([x_pos[0] - 1, x_pos[-1] + 0.2], [y_sft, y_sft], color='black', alpha=0.3, linewidth=4, linestyle='--', label='SFT only')
    # Ours with 25% data horizontal line.
    ax.plot([x_pos[0] - 0.1, x_pos[-1] + 0.1], [results_data_fraction[-1][1], results_data_fraction[-1][1]], color=colors[-1], linewidth=3, linestyle=':')
    for m, (method, color) in enumerate(zip(methods, colors)):
        y = results_data_fraction[m]
        ax.plot(x_pos, y, color=color, linewidth=3, marker='o', markersize=10, label=method)

    ax.set_xlabel(r'Fraction of data used for post-training', fontsize=28, labelpad=18)
    ax.set_xlim(x_pos[0] - 0.5, x_pos[-1] + 0.4)
    ax.set_xticks(x_pos)
    ax.set_xticklabels([f'{item:.0%}' for item in data_fraction_vals])
    ax.set_ylabel('$POPE_{Adv}$' + r'$\uparrow$', fontsize=28, fontfamily='monospace', labelpad=12)
    ax.set_yticks(y_ticks)
    ax.set_yticklabels(y_ticks)
    ax.tick_params(labelsize=24, length=8, width=1.5)
    ax.legend(fontsize=24, loc='lower center', ncols=2, frameon=False)

    ax = axes[1]
    y_ticks = [75, 79, 84, 88]
    beta_key = '$\beta$'
    beta_vals = np.asarray(data_ablation[beta_key])
    results_beta = data_ablation['results'][beta_key]  # shape (n_methods, n_beta)
    x_pos = np.arange(len(beta_vals))
    for m, (method, color) in enumerate(zip(methods, colors)):
        y = results_beta[m]
        ax.plot(x_pos, y, color=color, linewidth=3, marker='o', markersize=10, label=method)

    ax.set_xlabel(r'Hyperparameter $\beta$', fontsize=28, labelpad=18)
    ax.set_xticks(x_pos)
    ax.set_xticklabels([str(b) for b in beta_vals])
    ax.set_ylabel('$POPE_{Adv}$' + r'$\uparrow$', fontsize=28, fontfamily='monospace', labelpad=12)
    ax.set_yticks(y_ticks)
    ax.set_yticklabels(y_ticks)
    ax.tick_params(labelsize=24, length=8, width=1.5)
    ax.legend(fontsize=24, loc='lower center', frameon=False)

    ax = axes[2]
    y_ticks = [84, 85, 86, 87, 88]
    lambda_key = '$\lambda$'
    lambda_vals = np.asarray(data_ablation[lambda_key])
    results_lambda = data_ablation['results'][lambda_key]  # shape (2, n_beta)
    assert len(results_lambda) == 2
    x_pos = np.arange(len(lambda_vals))
    y = results_lambda[0]
    ax.plot(x_pos, y, color=colors[-1], linewidth=3, marker='o', markersize=10, label=methods[-1], alpha=0.4)
    ax.set_xlabel(r'Hyperparameter $\lambda$', fontsize=28, labelpad=18)
    ax.set_xticks(x_pos)
    ax.set_xlim(x_pos[0] - 0.5, x_pos[-1] + 0.5)
    ax.set_xticklabels([str(b) for b in lambda_vals])
    ax.set_yticks(y_ticks)
    ax.set_yticklabels(y_ticks)
    ax.set_ylabel('$POPE_{Adv}$' + r'$\uparrow$', fontsize=28, fontfamily='monospace', labelpad=12, color=colors[-1], alpha=0.4)
    ax.tick_params(labelsize=24, length=8, width=1.5)
    ax.legend(fontsize=24, loc='lower center', frameon=False)

    ax2 = ax.twinx()
    ax2.spines['right'].set_visible(True)
    y = results_lambda[1]
    ax2.set_ylabel('$MathVista$' + r'$\uparrow$', fontsize=28, fontfamily='monospace', labelpad=36, rotation=270, color=colors[-1], alpha=1.0)
    ax2.plot(x_pos, y, color=colors[-1], linewidth=3, marker='o', markersize=10, label=methods[-1], alpha=1.0)
    ax2.plot(x_pos, y, color=colors[-1], linewidth=3, marker='o', markersize=10, label=methods[-1], alpha=1.0)
    ax2.set_yticks([48, 48.5, 49, 49.5, 50])
    ax2.set_yticklabels([48, 48.5, 49, 49.5, 50])
    ax2.tick_params(labelsize=24, length=8, width=1.5)
    ax2.legend(fontsize=24, loc='lower center', frameon=False)

    # Right panel: at each x, print average of the two metrics at the top
    y_top = ax.get_ylim()[1] - 0.2
    for i in range(len(x_pos)):
        avg = (results_lambda[0][i] + results_lambda[1][i]) / 2
        ax.text(x_pos[i], y_top, f'Mean = {avg:.1f}', ha='center', va='bottom', fontsize=18)

    fig.tight_layout(pad=0.5)
    fig.subplots_adjust(wspace=0.3)
    os.makedirs('./figures/', exist_ok=True)
    fig.savefig('./figures/ablation_curves.png', dpi=300)
    plt.close(fig)
    return


if __name__ == '__main__':
    plt.rcParams['font.family'] = 'helvetica'
    plt.rcParams['font.size'] = 24
    plt.rcParams['axes.spines.top'] = False
    plt.rcParams['axes.spines.right'] = False
    plt.rcParams['axes.linewidth'] = 3

    plot_curves(data_ablation)
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 194kUpdated Claude CodeCodexVerifiedInstall