update

README.md

@@ -7,6 +7,19 @@ sdk: gradio
 sdk_version: 3.19.1
 app_file: app.py
 pinned: false
+tags:
+- evaluate
+- metric
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Metric Card for Top-5 error rate
+
+## Metric Description
+
+The "top-5 error" is the percentage of times that the target label does not appear among the 5 highest-probability predictions. It can be computed with:
+Top-5 Error Rate = 1 - Top-5 Accuracy
+or equivalently:
+Top-5 Error Rate = (Number of incorrect top-5 predictions) / (Total number of cases processed)
+ Where:
+- Top-5 Accuracy: The proportion of cases where the true label is among the model's top 5 predicted classes.
+- Incorrect top-5 prediction: The true label is not in the top 5 predicted classes (ranked by probability).

top5_error_rate.py

@@ -2,7 +2,6 @@ from typing import Dict, Any
 
 import datasets
 import evaluate
-import numpy as np
 from evaluate.utils.file_utils import add_start_docstrings
 
 _DESCRIPTION = """
@@ -43,9 +42,16 @@ class Top5ErrorRate(evaluate.Metric):
             inputs_description=_KWARGS_DESCRIPTION,
             features=datasets.Features(
                 {
-                    "predictions": datasets.Sequence(datasets.Value("int32")),
+                    "predictions": datasets.Sequence(
+                        datasets.Sequence(datasets.Value("int32"))
+                    ),
                     "references": datasets.Sequence(datasets.Value("int32")),
                 }
+                if self.config_name == "multilabel"
+                else {
+                    "predictions": datasets.Sequence(datasets.Value("int32")),
+                    "references": datasets.Value("int32"),
+                }
             ),
             reference_urls=[],
         )
@@ -57,7 +63,6 @@ class Top5ErrorRate(evaluate.Metric):
         references: list[int] = None,
         **kwargs,
     ) -> Dict[str, Any]:
-
         total = len(references)
         correct = sum(1 for pred, ref in zip(predictions, references) if ref in pred)