update

app.py

@@ -3,7 +3,7 @@ from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns
 import pandas as pd
 from apscheduler.schedulers.background import BackgroundScheduler
 from huggingface_hub import snapshot_download
-
+import plotly.graph_objects as go
 from src.about import (
     CITATION_BUTTON_LABEL,
     CITATION_BUTTON_TEXT,
@@ -58,6 +58,34 @@ LEADERBOARD_DF = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS,
     pending_eval_queue_df,
 ) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS)
 
+
+def init_perf_plot(df):
+    df = df.copy()
+    params_col = 'num_params'
+    df["symbol"] = 2  # Triangle
+    df["color"] = ""
+    df.loc[df["model"].str.contains("granite"), "color"] = "grey"
+    acc_col = 'failuresensor_mcqa_el'
+    fig = go.Figure()
+    for i in df.index:
+        fig.add_trace(
+            go.Scatter(
+                x=[df.loc[i, params_col]],
+                y=[df.loc[i, acc_col]],
+                name=df.loc[i, "model"]
+            )
+        )
+
+    fig.update_layout(
+        autosize=False,
+        width=650,
+        height=600,
+        title=f"Model Size Vs Accuracy",
+        xaxis_title=f"{params_col}",
+        yaxis_title="Accuracy",
+    )
+    return fig
+
 def init_leaderboard(dataframe):
     if dataframe is None or dataframe.empty:
         raise ValueError("Leaderboard DataFrame is empty or None.")
@@ -100,14 +128,14 @@ with demo:
             leaderboard = init_leaderboard(LEADERBOARD_DF)
 
         with gr.TabItem("📊 Performance Plot", elem_id="llm-benchmark-tab-table", id=1):
-            gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
-        print(LEADERBOARD_DF)
-            # with gr.Row():
-            #     bs_1_plot = gr.components.Plot(
-            #         value=plot_throughput(LEADERBOARD_DF, bs=1),
-            #         elem_id="bs1-plot",
-            #         show_label=False,
-            #     )
+            print(LEADERBOARD_DF)
+            # gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
+            with gr.Row():
+                bs_1_plot = gr.components.Plot(
+                    value=init_perf_plot(LEADERBOARD_DF, bs=1),
+                    elem_id="bs1-plot",
+                    show_label=False,
+                )
 
         with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
             gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")