Merged

app.py

@@ -4,6 +4,7 @@ import gradio as gr
 import pandas as pd
 from apscheduler.schedulers.background import BackgroundScheduler
 from huggingface_hub import snapshot_download
+import time
 
 from src.about import (
     CITATION_BUTTON_LABEL,
@@ -96,30 +97,38 @@ except Exception:
 # Span based results
 # changes to be made here
 
+start_time = time.time()
+
 _, harness_datasets_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, DATASET_COLS, DATASET_BENCHMARK_COLS, "accuracy", "datasets")
 harness_datasets_leaderboard_df = harness_datasets_original_df.copy()
+print("Closed ended English results loaded")
 
 _, open_ended_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, OPEN_ENDED_COLS, OPEN_ENDED_BENCHMARK_COLS, "score", "open_ended")
 open_ended_leaderboard_df = open_ended_original_df.copy()
+print("Open ended English results loaded")
 
 _, med_safety_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, MED_SAFETY_COLS, MED_SAFETY_BENCHMARK_COLS, "score", "med_safety")
 med_safety_leaderboard_df = med_safety_original_df.copy()
+print("Med safety results loaded")
 
 _, medical_summarization_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, MEDICAL_SUMMARIZATION_COLS, MEDICAL_SUMMARIZATION_BENCHMARK_COLS, "score", "medical_summarization")
 medical_summarization_leaderboard_df = medical_summarization_original_df.copy()
+print("Medical summarization results loaded")
 
 _, aci_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, ACI_COLS, ACI_BENCHMARK_COLS, "score", "aci")
 aci_leaderboard_df = aci_original_df.copy()
+print("ACI results loaded")
 
 _, soap_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, SOAP_COLS, SOAP_BENCHMARK_COLS, "score", "soap")
 soap_leaderboard_df = soap_original_df.copy()
+print("SOAP results loaded")
 
 _, healthbench_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, HEALTHBENCH_COLS, HEALTHBENCH_BENCHMARK_COLS, "score", "healthbench")
 healthbench_leaderboard_df = healthbench_original_df.copy()
 
 _, healthbench_hard_original_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, HEALTHBENCH_HARD_COLS, HEALTHBENCH_HARD_BENCHMARK_COLS, "score", "healthbench_hard")
 healthbench_hard_leaderboard_df = healthbench_hard_original_df.copy()
-
+print("Healthbench results loaded")
 
 _, open_ended_arabic_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, OpenEndedArabic_COLS, OpenEndedArabic_BENCHMARK_COLS, "score", "open_ended_arabic")
 _, open_ended_french_df = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, OpenEndedFrench_COLS, OpenEndedFrench_BENCHMARK_COLS, "score", "open_ended_french")
@@ -136,8 +145,14 @@ open_ended_portuguese_leaderboard_df = open_ended_portuguese_df.copy()
 open_ended_romanian_leaderboard_df = open_ended_romanian_df.copy()
 open_ended_greek_leaderboard_df = open_ended_greek_df.copy()
 open_ended_spanish_leaderboard_df = open_ended_spanish_df.copy()
+print("Open ended multilingual results loaded")
+
 closed_ended_multilingual_leaderboard_df = closed_ended_multilingual_df.copy()
+print("Closed ended multilingual results loaded")
 
+end_time = time.time()
+total_time = end_time - start_time
+print(f"Total time taken to load all results: {total_time:.2f} seconds")
 
 # breakpoint()
 # # Token based results
@@ -154,7 +169,7 @@ closed_ended_multilingual_leaderboard_df = closed_ended_multilingual_df.copy()
     pending_eval_queue_df,
 ) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS)
 
-breakpoint()
+# breakpoint()
 def update_df(shown_columns, subset="datasets"):
     # changes to be made here
     if subset == "datasets":
@@ -1323,117 +1338,6 @@ with demo:
                             queue=True,
                         )
     
-            with gr.Row():
-                with gr.Column():
-                    with gr.Row():
-                        search_bar = gr.Textbox(
-                            placeholder=" 🔍 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 and (c.invariant or c.dataset_task_col)],
-                            value=[
-                                c.name
-                                for c in fields(AutoEvalColumn)
-                                if c.displayed_by_default and not c.hidden and not c.never_hidden and (c.invariant or c.dataset_task_col)
-                            ],
-                            label="Select columns to show",
-                            elem_id="column-select",
-                            interactive=True,
-                        )
-                    # with gr.Row():
-                    #     deleted_models_visibility = gr.Checkbox(
-                    #         value=False, label="Show gated/private/deleted models", 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_architecture = gr.CheckboxGroup(
-                    #     label="Architecture Types",
-                    #     choices=[i.value.name for i in ModelArch],
-                    #     value=[i.value.name for i in ModelArch],
-                    #     interactive=True,
-                    #     elem_id="filter-columns-architecture",
-                    # )
-                    filter_domain_specific = gr.CheckboxGroup(
-                        label="Domain Specificity",
-                        choices=["🏥  Clinical models", "Generic models"],
-                        value=["🏥  Clinical models", "Generic models"],
-                        interactive=True,
-                        elem_id="filter-columns-type",
-                    )
-                    filter_columns_size = gr.CheckboxGroup(
-                        label="Model sizes (in billions of parameters)",
-                        choices=list(NUMERIC_INTERVALS.keys()),
-                        value=list(NUMERIC_INTERVALS.keys()),
-                        interactive=True,
-                        elem_id="filter-columns-size",
-                    )
-
-            datasets_leaderboard_df, datasets_original_df = update_df(shown_columns.value, subset="datasets")
-
-            leaderboard_table = gr.components.Dataframe(
-                value=datasets_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=datasets_original_df[DATASET_COLS],
-                headers=DATASET_COLS,
-                datatype=TYPES,
-                visible=False,
-            )
-
-                        
-            search_bar.submit(
-                update_table,
-                [
-                    hidden_leaderboard_table_for_search,
-                    shown_columns,
-                    search_bar,
-                    filter_columns_type,
-                    filter_domain_specific,
-                    filter_columns_size
-                    # filter_columns_architecture
-                ],
-                leaderboard_table,
-            )
-            for selector in [
-                shown_columns,
-                filter_columns_type,
-                filter_domain_specific,
-                # filter_columns_architecture,
-                filter_columns_size,
-                # deleted_models_visibility,
-            ]:
-                selector.change(
-                    update_table,
-                    [
-                        hidden_leaderboard_table_for_search,
-                        shown_columns,
-                        search_bar,
-                        filter_columns_type,
-                        filter_domain_specific,
-                        filter_columns_size
-                        # filter_columns_architecture,
-                    ],
-                    leaderboard_table,
-                    queue=True,
-                )
-
         with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=7):
             gr.Markdown(LLM_BENCHMARKS_TEXT_1, elem_classes="markdown-text")
             gr.HTML(FIVE_PILLAR_DIAGRAM)