Upload folder using huggingface_hub

README.md

@@ -1,12 +1,12 @@
+
 ---
-title: Mini Leaderboard Main
-emoji: 📉
-colorFrom: blue
-colorTo: pink
+title: mini_leaderboard_main 
+emoji: 🔥
+colorFrom: indigo
+colorTo: indigo
 sdk: gradio
 sdk_version: 4.25.0
-app_file: app.py
+app_file: run.py
 pinned: false
+hf_oauth: true
 ---
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

assets/custom_css.css

@@ -0,0 +1,87 @@
+/* Hides the final AutoEvalColumn */
+#llm-benchmark-tab-table table td:last-child,
+#llm-benchmark-tab-table table th:last-child {
+    display: none;
+}
+
+/* Limit the width of the first AutoEvalColumn so that names don't expand too much */
+table td:first-child,
+table th:first-child {
+    max-width: 400px;
+    overflow: auto;
+    white-space: nowrap;
+}
+
+/* Full width space */
+.gradio-container {
+  max-width: 95%!important;
+}
+
+/* Text style and margins */
+.markdown-text {
+    font-size: 16px !important;
+}
+
+#models-to-add-text {
+    font-size: 18px !important;
+}
+
+#citation-button span {
+    font-size: 16px !important;
+}
+
+#citation-button textarea {
+    font-size: 16px !important;
+}
+
+#citation-button > label > button {
+    margin: 6px;
+    transform: scale(1.3);
+}
+
+#search-bar-table-box > div:first-child {
+    background: none;
+    border: none;
+}
+ 
+#search-bar {
+    padding: 0px;
+}
+
+.tab-buttons button {
+    font-size: 20px;
+}
+
+/* Filters style */
+#filter_type{
+    border: 0;
+    padding-left: 0;
+    padding-top: 0;
+}
+#filter_type label {
+    display: flex;
+}
+#filter_type label > span{
+    margin-top: var(--spacing-lg);
+    margin-right: 0.5em;
+}
+#filter_type label > .wrap{
+    width: 103px;
+}
+#filter_type label > .wrap .wrap-inner{  
+    padding: 2px;
+}
+#filter_type label > .wrap .wrap-inner input{
+    width: 1px
+}
+#filter-columns-type{
+    border:0;
+    padding:0.5;
+}
+#filter-columns-size{
+    border:0;
+    padding:0.5;
+}
+#box-filter > .form{
+    border: 0
+}

requirements.txt

@@ -0,0 +1,2 @@
+gradio-client @ git+https://github.com/gradio-app/gradio@efd95245081f5657b1d13f34038205fc8791c1f7#subdirectory=client/python
+https://gradio-builds.s3.amazonaws.com/efd95245081f5657b1d13f34038205fc8791c1f7/gradio-4.25.0-py3-none-any.whl

run.ipynb

@@ -0,0 +1 @@
+{"cells": [{"cell_type": "markdown", "id": "302934307671667531413257853548643485645", "metadata": {}, "source": ["# Gradio Demo: mini_leaderboard"]}, {"cell_type": "code", "execution_count": null, "id": "272996653310673477252411125948039410165", "metadata": {}, "outputs": [], "source": ["!pip install -q gradio "]}, {"cell_type": "code", "execution_count": null, "id": "288918539441861185822528903084949547379", "metadata": {}, "outputs": [], "source": ["# Downloading files from the demo repo\n", "import os\n", "os.mkdir('assets')\n", "!wget -q -O assets/__init__.py https://github.com/gradio-app/gradio/raw/main/demo/mini_leaderboard/assets/__init__.py\n", "!wget -q -O assets/custom_css.css https://github.com/gradio-app/gradio/raw/main/demo/mini_leaderboard/assets/custom_css.css\n", "!wget -q -O assets/leaderboard_data.json https://github.com/gradio-app/gradio/raw/main/demo/mini_leaderboard/assets/leaderboard_data.json"]}, {"cell_type": "code", "execution_count": null, "id": "44380577570523278879349135829904343037", "metadata": {}, "outputs": [], "source": ["import gradio as gr\n", "import pandas as pd\n", "from pathlib import Path\n", "\n", "abs_path = Path(__file__).parent.absolute()\n", "\n", "df = pd.read_json(str(abs_path / \"assets/leaderboard_data.json\"))\n", "invisible_df = df.copy()\n", "\n", "\n", "COLS = [\n", "    \"T\",\n", "    \"Model\",\n", "    \"Average \u2b06\ufe0f\",\n", "    \"ARC\",\n", "    \"HellaSwag\",\n", "    \"MMLU\",\n", "    \"TruthfulQA\",\n", "    \"Winogrande\",\n", "    \"GSM8K\",\n", "    \"Type\",\n", "    \"Architecture\",\n", "    \"Precision\",\n", "    \"Merged\",\n", "    \"Hub License\",\n", "    \"#Params (B)\",\n", "    \"Hub \u2764\ufe0f\",\n", "    \"Model sha\",\n", "    \"model_name_for_query\",\n", "]\n", "ON_LOAD_COLS = [\n", "    \"T\",\n", "    \"Model\",\n", "    \"Average \u2b06\ufe0f\",\n", "    \"ARC\",\n", "    \"HellaSwag\",\n", "    \"MMLU\",\n", "    \"TruthfulQA\",\n", "    \"Winogrande\",\n", "    \"GSM8K\",\n", "    \"model_name_for_query\",\n", "]\n", "TYPES = [\n", "    \"str\",\n", "    \"markdown\",\n", "    \"number\",\n", "    \"number\",\n", "    \"number\",\n", "    \"number\",\n", "    \"number\",\n", "    \"number\",\n", "    \"number\",\n", "    \"str\",\n", "    \"str\",\n", "    \"str\",\n", "    \"str\",\n", "    \"bool\",\n", "    \"str\",\n", "    \"number\",\n", "    \"number\",\n", "    \"bool\",\n", "    \"str\",\n", "    \"bool\",\n", "    \"bool\",\n", "    \"str\",\n", "]\n", "NUMERIC_INTERVALS = {\n", "    \"?\": pd.Interval(-1, 0, closed=\"right\"),\n", "    \"~1.5\": pd.Interval(0, 2, closed=\"right\"),\n", "    \"~3\": pd.Interval(2, 4, closed=\"right\"),\n", "    \"~7\": pd.Interval(4, 9, closed=\"right\"),\n", "    \"~13\": pd.Interval(9, 20, closed=\"right\"),\n", "    \"~35\": pd.Interval(20, 45, closed=\"right\"),\n", "    \"~60\": pd.Interval(45, 70, closed=\"right\"),\n", "    \"70+\": pd.Interval(70, 10000, closed=\"right\"),\n", "}\n", "MODEL_TYPE = [str(s) for s in df[\"T\"].unique()]\n", "Precision = [str(s) for s in df[\"Precision\"].unique()]\n", "\n", "\n", "# Searching and filtering\n", "def update_table(\n", "    hidden_df: pd.DataFrame,\n", "    columns: list,\n", "    type_query: list,\n", "    precision_query: str,\n", "    size_query: list,\n", "    query: str,\n", "):\n", "    filtered_df = filter_models(hidden_df, type_query, size_query, precision_query)\n", "    filtered_df = filter_queries(query, filtered_df)\n", "    df = select_columns(filtered_df, columns)\n", "    return df\n", "\n", "\n", "def search_table(df: pd.DataFrame, query: str) -> pd.DataFrame:\n", "    return df[(df[\"model_name_for_query\"].str.contains(query, case=False))]\n", "\n", "\n", "def select_columns(df: pd.DataFrame, columns: list) -> pd.DataFrame:\n", "    # We use COLS to maintain sorting\n", "    filtered_df = df[[c for c in COLS if c in df.columns and c in columns]]\n", "    return filtered_df\n", "\n", "\n", "def filter_queries(query: str, filtered_df: pd.DataFrame) -> pd.DataFrame:\n", "    final_df = []\n", "    if query != \"\":\n", "        queries = [q.strip() for q in query.split(\";\")]\n", "        for _q in queries:\n", "            _q = _q.strip()\n", "            if _q != \"\":\n", "                temp_filtered_df = search_table(filtered_df, _q)\n", "                if len(temp_filtered_df) > 0:\n", "                    final_df.append(temp_filtered_df)\n", "        if len(final_df) > 0:\n", "            filtered_df = pd.concat(final_df)\n", "            filtered_df = filtered_df.drop_duplicates(\n", "                subset=[\"Model\", \"Precision\", \"Model sha\"]\n", "            )\n", "\n", "    return filtered_df\n", "\n", "\n", "def filter_models(\n", "    df: pd.DataFrame,\n", "    type_query: list,\n", "    size_query: list,\n", "    precision_query: list,\n", ") -> pd.DataFrame:\n", "    # Show all models\n", "    filtered_df = df\n", "\n", "    type_emoji = [t[0] for t in type_query]\n", "    filtered_df = filtered_df.loc[df[\"T\"].isin(type_emoji)]\n", "    filtered_df = filtered_df.loc[df[\"Precision\"].isin(precision_query + [\"None\"])]\n", "\n", "    numeric_interval = pd.IntervalIndex(\n", "        sorted([NUMERIC_INTERVALS[s] for s in size_query])\n", "    )\n", "    params_column = pd.to_numeric(df[\"#Params (B)\"], errors=\"coerce\")\n", "    mask = params_column.apply(lambda x: any(numeric_interval.contains(x)))\n", "    filtered_df = filtered_df.loc[mask]\n", "\n", "    return filtered_df\n", "\n", "\n", "demo = gr.Blocks(css=str(abs_path / \"assets/leaderboard_data.json\"))\n", "with demo:\n", "    gr.Markdown(\"\"\"Test Space of the LLM Leaderboard\"\"\", elem_classes=\"markdown-text\")\n", "\n", "    with gr.Tabs(elem_classes=\"tab-buttons\") as tabs:\n", "        with gr.TabItem(\"\ud83c\udfc5 LLM Benchmark\", elem_id=\"llm-benchmark-tab-table\", id=0):\n", "            with gr.Row():\n", "                with gr.Column():\n", "                    with gr.Row():\n", "                        search_bar = gr.Textbox(\n", "                            placeholder=\" \ud83d\udd0d Search for your model (separate multiple queries with `;`) and press ENTER...\",\n", "                            show_label=False,\n", "                            elem_id=\"search-bar\",\n", "                        )\n", "                    with gr.Row():\n", "                        shown_columns = gr.CheckboxGroup(\n", "                            choices=COLS,\n", "                            value=ON_LOAD_COLS,\n", "                            label=\"Select columns to show\",\n", "                            elem_id=\"column-select\",\n", "                            interactive=True,\n", "                        )\n", "                with gr.Column(min_width=320):\n", "                    filter_columns_type = gr.CheckboxGroup(\n", "                        label=\"Model types\",\n", "                        choices=MODEL_TYPE,\n", "                        value=MODEL_TYPE,\n", "                        interactive=True,\n", "                        elem_id=\"filter-columns-type\",\n", "                    )\n", "                    filter_columns_precision = gr.CheckboxGroup(\n", "                        label=\"Precision\",\n", "                        choices=Precision,\n", "                        value=Precision,\n", "                        interactive=True,\n", "                        elem_id=\"filter-columns-precision\",\n", "                    )\n", "                    filter_columns_size = gr.CheckboxGroup(\n", "                        label=\"Model sizes (in billions of parameters)\",\n", "                        choices=list(NUMERIC_INTERVALS.keys()),\n", "                        value=list(NUMERIC_INTERVALS.keys()),\n", "                        interactive=True,\n", "                        elem_id=\"filter-columns-size\",\n", "                    )\n", "\n", "            leaderboard_table = gr.components.Dataframe(\n", "                value=df[ON_LOAD_COLS],\n", "                headers=ON_LOAD_COLS,\n", "                datatype=TYPES,\n", "                elem_id=\"leaderboard-table\",\n", "                interactive=False,\n", "                visible=True,\n", "                column_widths=[\"2%\", \"33%\"],\n", "            )\n", "\n", "            # Dummy leaderboard for handling the case when the user uses backspace key\n", "            hidden_leaderboard_table_for_search = gr.components.Dataframe(\n", "                value=invisible_df[COLS],\n", "                headers=COLS,\n", "                datatype=TYPES,\n", "                visible=False,\n", "            )\n", "            search_bar.submit(\n", "                update_table,\n", "                [\n", "                    hidden_leaderboard_table_for_search,\n", "                    shown_columns,\n", "                    filter_columns_type,\n", "                    filter_columns_precision,\n", "                    filter_columns_size,\n", "                    search_bar,\n", "                ],\n", "                leaderboard_table,\n", "            )\n", "            for selector in [\n", "                shown_columns,\n", "                filter_columns_type,\n", "                filter_columns_precision,\n", "                filter_columns_size,\n", "            ]:\n", "                selector.change(\n", "                    update_table,\n", "                    [\n", "                        hidden_leaderboard_table_for_search,\n", "                        shown_columns,\n", "                        filter_columns_type,\n", "                        filter_columns_precision,\n", "                        filter_columns_size,\n", "                        search_bar,\n", "                    ],\n", "                    leaderboard_table,\n", "                    queue=True,\n", "                )\n", "\n", "\n", "if __name__ == \"__main__\":\n", "    demo.queue(default_concurrency_limit=40).launch()\n"]}], "metadata": {}, "nbformat": 4, "nbformat_minor": 5}

run.py

@@ -0,0 +1,244 @@
+import gradio as gr
+import pandas as pd
+from pathlib import Path
+
+abs_path = Path(__file__).parent.absolute()
+
+df = pd.read_json(str(abs_path / "assets/leaderboard_data.json"))
+invisible_df = df.copy()
+
+
+COLS = [
+    "T",
+    "Model",
+    "Average ⬆️",
+    "ARC",
+    "HellaSwag",
+    "MMLU",
+    "TruthfulQA",
+    "Winogrande",
+    "GSM8K",
+    "Type",
+    "Architecture",
+    "Precision",
+    "Merged",
+    "Hub License",
+    "#Params (B)",
+    "Hub ❤️",
+    "Model sha",
+    "model_name_for_query",
+]
+ON_LOAD_COLS = [
+    "T",
+    "Model",
+    "Average ⬆️",
+    "ARC",
+    "HellaSwag",
+    "MMLU",
+    "TruthfulQA",
+    "Winogrande",
+    "GSM8K",
+    "model_name_for_query",
+]
+TYPES = [
+    "str",
+    "markdown",
+    "number",
+    "number",
+    "number",
+    "number",
+    "number",
+    "number",
+    "number",
+    "str",
+    "str",
+    "str",
+    "str",
+    "bool",
+    "str",
+    "number",
+    "number",
+    "bool",
+    "str",
+    "bool",
+    "bool",
+    "str",
+]
+NUMERIC_INTERVALS = {
+    "?": pd.Interval(-1, 0, closed="right"),
+    "~1.5": pd.Interval(0, 2, closed="right"),
+    "~3": pd.Interval(2, 4, closed="right"),
+    "~7": pd.Interval(4, 9, closed="right"),
+    "~13": pd.Interval(9, 20, closed="right"),
+    "~35": pd.Interval(20, 45, closed="right"),
+    "~60": pd.Interval(45, 70, closed="right"),
+    "70+": pd.Interval(70, 10000, closed="right"),
+}
+MODEL_TYPE = [str(s) for s in df["T"].unique()]
+Precision = [str(s) for s in df["Precision"].unique()]
+
+
+# Searching and filtering
+def update_table(
+    hidden_df: pd.DataFrame,
+    columns: list,
+    type_query: list,
+    precision_query: str,
+    size_query: list,
+    query: str,
+):
+    filtered_df = filter_models(hidden_df, type_query, size_query, precision_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["model_name_for_query"].str.contains(query, case=False))]
+
+
+def select_columns(df: pd.DataFrame, columns: list) -> pd.DataFrame:
+    # We use COLS to maintain sorting
+    filtered_df = df[[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=["Model", "Precision", "Model sha"]
+            )
+
+    return filtered_df
+
+
+def filter_models(
+    df: pd.DataFrame,
+    type_query: list,
+    size_query: list,
+    precision_query: list,
+) -> pd.DataFrame:
+    # Show all models
+    filtered_df = df
+
+    type_emoji = [t[0] for t in type_query]
+    filtered_df = filtered_df.loc[df["T"].isin(type_emoji)]
+    filtered_df = filtered_df.loc[df["Precision"].isin(precision_query + ["None"])]
+
+    numeric_interval = pd.IntervalIndex(
+        sorted([NUMERIC_INTERVALS[s] for s in size_query])
+    )
+    params_column = pd.to_numeric(df["#Params (B)"], errors="coerce")
+    mask = params_column.apply(lambda x: any(numeric_interval.contains(x)))
+    filtered_df = filtered_df.loc[mask]
+
+    return filtered_df
+
+
+demo = gr.Blocks(css=str(abs_path / "assets/leaderboard_data.json"))
+with demo:
+    gr.Markdown("""Test Space of the LLM Leaderboard""", elem_classes="markdown-text")
+
+    with gr.Tabs(elem_classes="tab-buttons") as tabs:
+        with gr.TabItem("🏅 LLM Benchmark", elem_id="llm-benchmark-tab-table", id=0):
+            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=COLS,
+                            value=ON_LOAD_COLS,
+                            label="Select columns to show",
+                            elem_id="column-select",
+                            interactive=True,
+                        )
+                with gr.Column(min_width=320):
+                    filter_columns_type = gr.CheckboxGroup(
+                        label="Model types",
+                        choices=MODEL_TYPE,
+                        value=MODEL_TYPE,
+                        interactive=True,
+                        elem_id="filter-columns-type",
+                    )
+                    filter_columns_precision = gr.CheckboxGroup(
+                        label="Precision",
+                        choices=Precision,
+                        value=Precision,
+                        interactive=True,
+                        elem_id="filter-columns-precision",
+                    )
+                    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",
+                    )
+
+            leaderboard_table = gr.components.Dataframe(
+                value=df[ON_LOAD_COLS],
+                headers=ON_LOAD_COLS,
+                datatype=TYPES,
+                elem_id="leaderboard-table",
+                interactive=False,
+                visible=True,
+                column_widths=["2%", "33%"],
+            )
+
+            # Dummy leaderboard for handling the case when the user uses backspace key
+            hidden_leaderboard_table_for_search = gr.components.Dataframe(
+                value=invisible_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_precision,
+                    filter_columns_size,
+                    search_bar,
+                ],
+                leaderboard_table,
+            )
+            for selector in [
+                shown_columns,
+                filter_columns_type,
+                filter_columns_precision,
+                filter_columns_size,
+            ]:
+                selector.change(
+                    update_table,
+                    [
+                        hidden_leaderboard_table_for_search,
+                        shown_columns,
+                        filter_columns_type,
+                        filter_columns_precision,
+                        filter_columns_size,
+                        search_bar,
+                    ],
+                    leaderboard_table,
+                    queue=True,
+                )
+
+
+if __name__ == "__main__":
+    demo.queue(default_concurrency_limit=40).launch()