import subprocess import gradio as gr 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, EVALUATION_QUEUE_TEXT, QUESTION_FORMAT_TEXT, MACRO_AREA_TEXT, CHANGELOG_TEXT, ) from src.display.css_html_js import custom_css from src.display.utils import ( BENCHMARK_COLS, COLS, EVAL_COLS, SIZE_INTERVALS, TYPES, AutoEvalColumn, ModelType, fields, ) 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 add_new_eval def restart_space(): API.restart_space(repo_id=REPO_ID) 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() raw_data, original_df = get_leaderboard_df( EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS) leaderboard_df = original_df.copy() ( finished_eval_queue_df, running_eval_queue_df, pending_eval_queue_df, ) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS) # Searching and filtering def update_table( hidden_df: pd.DataFrame, columns: list, type_query: list, size_query: list, query: str, ): filtered_df = filter_models( hidden_df, type_query, size_query) filtered_df = filter_queries(query, filtered_df) df = select_columns(filtered_df, columns) return df def search_table(df: pd.DataFrame, query: str) -> pd.DataFrame: return df[(df[AutoEvalColumn.model.name].str.contains(query, case=False))] def select_columns(df: pd.DataFrame, columns: list) -> pd.DataFrame: always_here_cols = [ AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name, ] # We use COLS to maintain sorting filtered_df = df[ always_here_cols + [c for c in COLS if c in df.columns and c in columns] ] return filtered_df def filter_queries(query: str, filtered_df: pd.DataFrame) -> pd.DataFrame: final_df = [] if query != "": queries = [q.strip() for q in query.split(";")] for _q in queries: _q = _q.strip() if _q != "": temp_filtered_df = search_table(filtered_df, _q) if len(temp_filtered_df) > 0: final_df.append(temp_filtered_df) if len(final_df) > 0: filtered_df = pd.concat(final_df) filtered_df = filtered_df.drop_duplicates( subset=[AutoEvalColumn.model.name] ) return filtered_df def filter_models( df: pd.DataFrame, type_query: list, size_query: list, ) -> pd.DataFrame: filtered_df = df type_emoji = [t[0] for t in type_query] filtered_df = filtered_df.loc[df[AutoEvalColumn.model_type_symbol.name].isin( type_emoji)] filtered_df = filtered_df.loc[df[AutoEvalColumn.params.name].isin( size_query)] return filtered_df def get_macro_area_data(): dataset = pd.read_csv("src/macro_area.csv", sep=',', skiprows=1) dataset = dataset.iloc[1:] columns = ['Model', 'LI (108)', 'RM (179)', 'RC (33)', 'WF (7)', 'LS (29)', ' MO (24)', 'SP (4)', 'SY (19)', 'TP (6)'] dataset.columns = columns dataset = dataset.round(1) # dataset = dataset.style.highlight_max(color='lightgreen', axis=0) return dataset def get_question_format_data(): dataset = pd.read_csv("src/question_format.csv", sep=',') top_level_headers = [ int(float(x)) if '.' in x else x for x in dataset.columns] second_level_headers = dataset.iloc[0] columns = [f'{x} - {y}' for x, y in zip(top_level_headers, second_level_headers)] dataset = dataset.iloc[2:] columns[0] = 'Model' dataset.columns = columns dataset = dataset.round(1) # dataset = dataset.style.highlight_max(color='lightgreen', axis=0) return dataset 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("πŸ… LLM Benchmark", elem_id="llm-benchmark-tab-table", id=0): with gr.Row(): with gr.Column(): with gr.Row(): search_bar = gr.Textbox( placeholder=" πŸ” Search for your model (separate multiple queries with `;`) and press ENTER...", show_label=False, elem_id="search-bar", ) with gr.Row(): shown_columns = gr.CheckboxGroup( choices=[ c.name for c in fields(AutoEvalColumn) if not c.hidden and not c.never_hidden ], value=[ c.name for c in fields(AutoEvalColumn) if c.displayed_by_default and not c.hidden and not c.never_hidden ], label="Select columns to show", elem_id="column-select", interactive=True, ) with gr.Column(min_width=320): # with gr.Box(elem_id="box-filter"): filter_columns_type = gr.CheckboxGroup( label="Model types", choices=[t.to_str() for t in ModelType], value=[t.to_str() for t in ModelType], interactive=True, elem_id="filter-columns-type", ) filter_columns_size = gr.CheckboxGroup( label="Model sizes", choices=list(SIZE_INTERVALS), value=list(SIZE_INTERVALS), interactive=True, elem_id="filter-columns-size", ) leaderboard_table = gr.components.Dataframe( value=leaderboard_df[ [c.name for c in fields(AutoEvalColumn) if c.never_hidden] + shown_columns.value ], headers=[c.name for c in fields( AutoEvalColumn) if c.never_hidden] + shown_columns.value, datatype=TYPES, elem_id="leaderboard-table", interactive=False, visible=True, ) # Dummy leaderboard for handling the case when the user uses backspace key hidden_leaderboard_table_for_search = gr.components.Dataframe( value=original_df[COLS], headers=COLS, datatype=TYPES, visible=False, ) search_bar.submit( update_table, [ hidden_leaderboard_table_for_search, shown_columns, filter_columns_type, filter_columns_size, search_bar, ], leaderboard_table, ) for selector in [shown_columns, filter_columns_type, filter_columns_size]: selector.change( update_table, [ hidden_leaderboard_table_for_search, shown_columns, filter_columns_type, filter_columns_size, search_bar, ], leaderboard_table, queue=True, ) with gr.TabItem('πŸ” In Depth Evaluation'): gr.Markdown('''# In Depth Evaluation''') gr.Markdown(QUESTION_FORMAT_TEXT) gr.Dataframe(get_question_format_data()) with gr.TabItem('πŸ—ΊοΈ Evaluation by Macro Area'): gr.Markdown('''# Macro Area Evaluation''') gr.Markdown(MACRO_AREA_TEXT) gr.Dataframe(get_macro_area_data()) with gr.TabItem("πŸ“ About", elem_id="llm-benchmark-tab-table", id=2): gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text") with gr.TabItem("πŸ› οΈ Changelog", elem_id="changelog-tab", id=3): gr.Markdown(CHANGELOG_TEXT, elem_classes="markdown-text") with gr.TabItem("πŸš€ Submit here! ", elem_id="llm-benchmark-tab-table", id=4): with gr.Column(): with gr.Row(): gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text") # with gr.Column(): # with gr.Accordion( # f"βœ… Finished Evaluations ({len(finished_eval_queue_df)})", # open=False, # ): # with gr.Row(): # finished_eval_table = gr.components.Dataframe( # value=finished_eval_queue_df, # headers=EVAL_COLS, # datatype=EVAL_TYPES, # row_count=5, # ) # with gr.Accordion( # f"πŸ”„ Running Evaluation Queue ({len(running_eval_queue_df)})", # open=False, # ): # with gr.Row(): # running_eval_table = gr.components.Dataframe( # value=running_eval_queue_df, # headers=EVAL_COLS, # datatype=EVAL_TYPES, # row_count=5, # ) # with gr.Accordion( # f"⏳ Pending Evaluation Queue ({len(pending_eval_queue_df)})", # open=False, # ): # with gr.Row(): # pending_eval_table = gr.components.Dataframe( # value=pending_eval_queue_df, # headers=EVAL_COLS, # datatype=EVAL_TYPES, # row_count=5, # ) # with gr.Row(): # gr.Markdown("# βœ‰οΈβœ¨ Submit your model 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 commit", placeholder="main") # model_type = gr.Dropdown( # choices=[t.to_str(" : ") for t in ModelType if t != ModelType.Unknown], # label="Model type", # multiselect=False, # value=None, # interactive=True, # ) # with gr.Column(): # precision = gr.Dropdown( # choices=[i.value.name for i in Precision if i != Precision.Unknown], # label="Precision", # multiselect=False, # value="float16", # interactive=True, # ) # weight_type = gr.Dropdown( # choices=[i.value.name for i in WeightType], # label="Weights type", # multiselect=False, # value="Original", # interactive=True, # ) # base_model_name_textbox = gr.Textbox(label="Base model (for delta or adapter weights)") # submit_button = gr.Button("Submit Eval") # submission_result = gr.Markdown() # submit_button.click( # add_new_eval, # [ # model_name_textbox, # base_model_name_textbox, # revision_name_textbox, # precision, # weight_type, # model_type, # ], # submission_result, # ) with gr.Row(): with gr.Accordion("πŸ“™ 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()