Use color map for Results metrics values

src/results.py

@@ -1,7 +1,6 @@
 import asyncio
 
 import gradio as gr
-import numpy as np
 import pandas as pd
 from huggingface_hub import HfFileSystem
 
@@ -67,6 +66,7 @@ def display_tab(tab, df, task, hide_errors=True, show_only_differences=False):
     if show_only_differences:
         any_difference = df.ne(df.iloc[:, 0], axis=0).any(axis=1)
     df = df.style.format(escape="html", na_rep="")
+    # Hide rows
     df.hide(
         [
             row
@@ -88,7 +88,18 @@ def display_tab(tab, df, task, hide_errors=True, show_only_differences=False):
         ],
         axis="index",
     )
-    df.apply(highlight_min_max, axis=1)
+    # Color metric result cells
+    idx = pd.IndexSlice
+    colored_rows = idx[
+        [
+            row
+            for row in df.index
+            if row.endswith("acc,none") or row.endswith("acc_norm,none") or row.endswith("exact_match,none")
+        ]
+    ]  # Apply only on numeric cells, otherwise the background gradient will not work
+    subset = idx[colored_rows, idx[:]]
+    df.background_gradient(cmap="Greens", vmin=0, vmax=1, subset=subset, axis=None)
+    # Format index values: remove prefix and suffix
     start = len(f"{tab}.leaderboard_") if task == "All" else len(f"{tab}.{task} ")
     df.format_index(lambda idx: idx[start:].removesuffix(",none"), axis="index")
     return df.to_html()
@@ -127,12 +138,5 @@ def clear_results():
     )
 
 
-def highlight_min_max(s):
-    if s.name.endswith("acc,none") or s.name.endswith("acc_norm,none") or s.name.endswith("exact_match,none"):
-        return np.where(s == np.nanmax(s.values), "background-color:green", "background-color:#D81B60")
-    else:
-        return [""] * len(s)
-
-
 def display_loading_message_for_results():
     return ("<h3 style='text-align: center;'>Loading...</h3>",) * 2