import gradio as gr from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns import pandas as pd from apscheduler.schedulers.background import BackgroundScheduler from huggingface_hub import snapshot_download from src.about import ( CITATION_BUTTON_LABEL, CITATION_BUTTON_TEXT, EVALUATION_QUEUE_TEXT, INTRODUCTION_TEXT, LLM_BENCHMARKS_TEXT, TITLE, ) from src.display.css_html_js import custom_css from src.display.utils import ( BENCHMARK_COLS, SPEECH_BENCHMARK_COLS, COLS, COLS_SPEECH, EVAL_COLS, EVAL_TYPES, AutoEvalColumn, AutoEvalColumnSpeech, ModelType, fields, WeightType, Precision, REGION_MAP ) from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN from src.populate import get_evaluation_queue_df, get_leaderboard_df from src.submission.submit import handle_csv_submission text_sample_path = "src/submission_samples/model_name_text.csv" speech_sample_path = "src/submission_samples/model_name_speech.csv" def restart_space(): API.restart_space(repo_id=REPO_ID) ### Space initialisation try: print(EVAL_REQUESTS_PATH) snapshot_download( repo_id=QUEUE_REPO, local_dir=EVAL_REQUESTS_PATH, repo_type="dataset", tqdm_class=None, etag_timeout=30, token=TOKEN ) except Exception: restart_space() try: print(EVAL_RESULTS_PATH) snapshot_download( repo_id=RESULTS_REPO, local_dir=EVAL_RESULTS_PATH, repo_type="dataset", tqdm_class=None, etag_timeout=30, token=TOKEN ) except Exception: restart_space() ( finished_eval_queue_df, running_eval_queue_df, pending_eval_queue_df, ) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS) def init_leaderboard(dataframe, result_type='text'): if dataframe is None or dataframe.empty: raise ValueError("Leaderboard DataFrame is empty or None.") column_class = AutoEvalColumn if result_type == "text" else AutoEvalColumnSpeech return Leaderboard( value=dataframe, datatype=[c.type for c in fields(column_class)], select_columns=SelectColumns( default_selection=[c.name for c in fields(column_class) if c.displayed_by_default], cant_deselect=[c.name for c in fields(column_class) if c.never_hidden], label="Select Columns to Display:", ), # search_columns=[AutoEvalColumn.model.name, AutoEvalColumn.license.name], search_columns=[column_class.model.name], hide_columns=[c.name for c in fields(column_class) if c.hidden], filter_columns=[ # ColumnFilter(AutoEvalColumn.model_type.name, type="checkboxgroup", label="Model types"), # ColumnFilter(AutoEvalColumn.precision.name, type="checkboxgroup", label="Precision"), # ColumnFilter( # AutoEvalColumn.params.name, # type="slider", # min=0.01, # max=150, # label="Select the number of parameters (B)", # ), # ColumnFilter( # AutoEvalColumn.still_on_hub.name, type="boolean", label="Deleted/incomplete", default=True # ), ], bool_checkboxgroup_label="Hide models", interactive=False, ) leaderboard_dataframes = { region: get_leaderboard_df( EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS, region if region != "All" else None, result_type="text" ) for region in REGION_MAP.values() } leaderboard_dataframes_speech = { region: get_leaderboard_df( EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS_SPEECH, SPEECH_BENCHMARK_COLS, region if region != "All" else None, result_type="speech" ) for region in REGION_MAP.values() } # Preload leaderboard blocks js_switch_code = """ (displayRegion) => { const regionMap = { "All": "All", "Africa": "Africa", "Americas/Oceania": "Americas_Oceania", "Asia (S)": "Asia_S", "Asia (SE)": "Asia_SE", "Asia (W, C)": "Asia_W_C", "Asia (E)": "Asia_E", "Europe (W, N, S)": "Europe_W_N_S", "Europe (E)": "Europe_E" }; const region = regionMap[displayRegion]; document.querySelectorAll('[id^="leaderboard-"]').forEach(el => el.classList.remove("visible")); const target = document.getElementById("leaderboard-" + region); if (target) { target.classList.add("visible"); // 🧠 Trigger reflow to fix row cutoff void target.offsetHeight; // Trigger reflow target.style.display = "none"; // Hide momentarily requestAnimationFrame(() => { target.style.display = ""; }); } } """ demo = gr.Blocks(css=custom_css) with demo: gr.HTML(TITLE) gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text") with gr.Tabs(elem_classes="tab-buttons") as tabs: with gr.TabItem("🏅 mSTEB Text Benchmark", elem_id="llm-benchmark-tab-table", id=0): with gr.Row(): region_dropdown = gr.Dropdown( choices=list(REGION_MAP.keys()), label="Select Region", value="All", interactive=True, ) # Region-specific leaderboard containers for display_name, region_key in REGION_MAP.items(): with gr.Column( elem_id=f"leaderboard-{region_key}", elem_classes=["visible"] if region_key == "All" else [] ): init_leaderboard(leaderboard_dataframes[region_key], result_type="text") # JS hook to toggle visible leaderboard region_dropdown.change(None, js=js_switch_code, inputs=[region_dropdown]) with gr.TabItem("🗣️ mSTEB Speech Benchmark", elem_id="speech-benchmark-tab-table", id=1): with gr.Row(): speech_region_dropdown = gr.Dropdown( choices=list(REGION_MAP.keys()), label="Select Region", value="All", interactive=True, ) for display_name, region_key in REGION_MAP.items(): with gr.Column( elem_id=f"speech-leaderboard-{region_key}", elem_classes=["visible"] if region_key == "All" else [] ): init_leaderboard(leaderboard_dataframes_speech[region_key], result_type='speech') speech_region_dropdown.change( None, js=js_switch_code.replace("leaderboard-", "speech-leaderboard-"), inputs=[speech_region_dropdown] ) with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2): gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text") with gr.TabItem("🚀 Submit here! ", elem_id="llm-benchmark-tab-table", id=3): with gr.Column(): with gr.Row(): gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text") with gr.Row(): gr.File( label="📄 Sample Text CSV", value=text_sample_path, interactive=False, file_types=[".csv"] ) gr.File( label="📄 Sample Speech CSV", value=speech_sample_path, interactive=False, file_types=[".csv"] ) with gr.Row(): gr.Markdown("# ✉️✨ Submit your model here!", elem_classes="markdown-text") with gr.Column(): model_name_textbox = gr.Textbox(label="Model name") result_type = gr.Radio(choices=["text", "speech"], label="Result Type", value="text") csv_file = gr.File(label="Upload CSV File", file_types=[".csv"]) submit_button = gr.Button("Submit Eval") submission_result = gr.Markdown() submit_button.click( handle_csv_submission, [ model_name_textbox, csv_file, result_type, ], submission_result, ) with gr.Row(): with gr.Accordion("📙 Citation", open=False): citation_button = gr.Textbox( value=CITATION_BUTTON_TEXT, label=CITATION_BUTTON_LABEL, lines=20, elem_id="citation-button", show_copy_button=True, ) scheduler = BackgroundScheduler() scheduler.add_job(restart_space, "interval", seconds=1800) scheduler.start() demo.queue(default_concurrency_limit=40).launch()