Spaces:
Sleeping
Sleeping
| __all__ = ['block', 'make_clickable_model', 'make_clickable_user', 'get_submissions'] | |
| import gradio as gr | |
| import pandas as pd | |
| import re | |
| import os | |
| import json | |
| import yaml | |
| import matplotlib.pyplot as plt | |
| from src.about import * | |
| from src.bin.PROBE import run_probe | |
| global data_component, filter_component | |
| def get_baseline_df(selected_methods, selected_metrics): | |
| df = pd.read_csv(CSV_RESULT_PATH) | |
| present_columns = ["method_name"] + selected_metrics | |
| df = df[df['method_name'].isin(selected_methods)][present_columns] | |
| return df | |
| # Function to create the plot | |
| def create_plot(methods_selected, x_metric, y_metric): | |
| df = pd.read_csv(CSV_RESULT_PATH) | |
| filtered_df = df[df['method_name'].isin(methods_selected)] | |
| # Create a larger plot | |
| plt.figure(figsize=(10, 8)) # Increase the figure size | |
| for method in methods_selected: | |
| method_data = filtered_df[filtered_df['method_name'] == method] | |
| plt.plot(method_data[x_metric], method_data[y_metric], label=method, marker='o') | |
| plt.xlabel(x_metric) | |
| plt.ylabel(y_metric) | |
| plt.title(f'{y_metric} vs {x_metric} for selected methods') | |
| plt.legend() | |
| plt.grid(True) | |
| # Save the plot to display it in Gradio | |
| plot_path = "plot.png" | |
| plt.savefig(plot_path) | |
| plt.close() | |
| return plot_path | |
| def add_new_eval( | |
| human_file, | |
| skempi_file, | |
| model_name_textbox: str, | |
| revision_name_textbox: str, | |
| benchmark_type, | |
| similarity_tasks, | |
| function_prediction_aspect, | |
| function_prediction_dataset, | |
| family_prediction_dataset, | |
| ): | |
| representation_name = model_name_textbox if revision_name_textbox == '' else revision_name_textbox | |
| results = run_probe(benchmark_type, representation_name, human_file, skempi_file, similarity_tasks, function_prediction_aspect, function_prediction_dataset, family_prediction_dataset) | |
| return None | |
| # Function to update leaderboard dynamically based on user selection | |
| def update_leaderboard(selected_methods, selected_metrics): | |
| return get_baseline_df(selected_methods, selected_metrics) | |
| block = gr.Blocks() | |
| with block: | |
| gr.Markdown(LEADERBOARD_INTRODUCTION) | |
| with gr.Tabs(elem_classes="tab-buttons") as tabs: | |
| # table jmmmu bench | |
| with gr.TabItem("🏅 PROBE Benchmark", elem_id="probe-benchmark-tab-table", id=1): | |
| method_names = pd.read_csv(CSV_RESULT_PATH)['method_name'].unique().tolist() | |
| metric_names = pd.read_csv(CSV_RESULT_PATH).columns.tolist() | |
| metrics_with_method = metric_names.copy() | |
| metric_names.remove('method_name') # Remove method_name from the metric options | |
| # Leaderboard section with method and metric selectors | |
| with gr.Row(): | |
| # Add method and metric selectors for leaderboard | |
| leaderboard_method_selector = gr.CheckboxGroup( | |
| choices=method_names, label="Select method_names for Leaderboard", value=method_names, interactive=True | |
| ) | |
| leaderboard_metric_selector = gr.CheckboxGroup( | |
| choices=metric_names, label="Select Metrics for Leaderboard", value=metric_names, interactive=True | |
| ) | |
| # Display the filtered leaderboard | |
| baseline_value = get_baseline_df(method_names, metric_names) | |
| baseline_header = ["method_name"] + metric_names | |
| baseline_datatype = ['markdown'] + ['number'] * len(metric_names) | |
| data_component = gr.components.Dataframe( | |
| value=baseline_value, | |
| headers=baseline_header, | |
| type="pandas", | |
| datatype=baseline_datatype, | |
| interactive=False, | |
| visible=True, | |
| ) | |
| # Update leaderboard when method/metric selection changes | |
| leaderboard_method_selector.change( | |
| update_leaderboard, | |
| inputs=[leaderboard_method_selector, leaderboard_metric_selector], | |
| outputs=data_component | |
| ) | |
| leaderboard_metric_selector.change( | |
| update_leaderboard, | |
| inputs=[leaderboard_method_selector, leaderboard_metric_selector], | |
| outputs=data_component | |
| ) | |
| # Add the visualizer components (Dropdown, Checkbox, Button, Image) | |
| with gr.Row(): | |
| # Visualizer Controls: Smaller and underneath each other | |
| with gr.Column(scale=1): | |
| method_selector = gr.CheckboxGroup(choices=method_names, label="Select method_names", interactive=True, value=method_names) | |
| x_metric_selector = gr.Dropdown(choices=metrics_with_method, label="Select X-axis Metric", interactive=True) | |
| y_metric_selector = gr.Dropdown(choices=metrics_with_method, label="Select Y-axis Metric", interactive=True) | |
| plot_button = gr.Button("Plot") | |
| # Larger plot display | |
| with gr.Column(scale=3): | |
| output_plot = gr.Image(label="Plot", height=735) # Set larger height for the plot | |
| plot_button.click(create_plot, inputs=[method_selector, x_metric_selector, y_metric_selector], outputs=output_plot) | |
| # table 5 | |
| with gr.TabItem("📝 About", elem_id="probe-benchmark-tab-table", id=2): | |
| with gr.Row(): | |
| gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text") | |
| with gr.TabItem("🚀 Submit here! ", elem_id="probe-benchmark-tab-table", id=3): | |
| with gr.Row(): | |
| gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text") | |
| with gr.Row(): | |
| gr.Markdown("# ✉️✨ Submit your model's representation files here!", elem_classes="markdown-text") | |
| with gr.Row(): | |
| with gr.Column(): | |
| model_name_textbox = gr.Textbox( | |
| label="Model name", | |
| ) | |
| revision_name_textbox = gr.Textbox( | |
| label="Revision Model Name", | |
| ) | |
| benchmark_type = gr.CheckboxGroup( | |
| choices=TASK_INFO, | |
| label="Benchmark Type", | |
| interactive=True, | |
| ) | |
| similarity_tasks = gr.CheckboxGroup( | |
| choices=similarity_tasks_options, | |
| label="Select Similarity Tasks", | |
| interactive=True, | |
| ) | |
| function_prediction_aspect = gr.Radio( | |
| choices=function_prediction_aspect_options, | |
| label="Select Function Prediction Aspect", | |
| interactive=True, | |
| ) | |
| function_prediction_dataset = gr.Radio( | |
| choices=function_prediction_dataset_options, | |
| label="Select Function Prediction Dataset", | |
| interactive=True, | |
| ) | |
| family_prediction_dataset = gr.CheckboxGroup( | |
| choices=family_prediction_dataset_options, | |
| label="Select Family Prediction Dataset", | |
| interactive=True, | |
| ) | |
| with gr.Column(): | |
| human_file = gr.components.File(label="Click to Upload the representation file (csv) for Human dataset", file_count="single", type='filepath') | |
| skempi_file = gr.components.File(label="Click to Upload the representation file (csv) for SKEMPI dataset", file_count="single", type='filepath') | |
| submit_button = gr.Button("Submit Eval") | |
| submission_result = gr.Markdown() | |
| submit_button.click( | |
| add_new_eval, | |
| inputs=[ | |
| human_file, | |
| skempi_file, | |
| model_name_textbox, | |
| revision_name_textbox, | |
| benchmark_type, | |
| similarity_tasks, | |
| function_prediction_aspect, | |
| function_prediction_dataset, | |
| family_prediction_dataset, | |
| ], | |
| ) | |
| def refresh_data(): | |
| value = get_baseline_df(method_names, metric_names) | |
| return value | |
| with gr.Row(): | |
| data_run = gr.Button("Refresh") | |
| data_run.click(refresh_data, outputs=[data_component]) | |
| with gr.Accordion("Citation", open=False): | |
| citation_button = gr.Textbox( | |
| value=CITATION_BUTTON_TEXT, | |
| label=CITATION_BUTTON_LABEL, | |
| elem_id="citation-button", | |
| show_copy_button=True, | |
| ) | |
| block.launch() | |