Update app.py

app.py

@@ -10,28 +10,34 @@ api = HfApi(token=token)
 
 
 def compute_df():
-    
+    api = HfApi()
     # download all files in https://huggingface.co/illuin-cde/baselines
-    files = [f for f in api.list_repo_files("illuin-cde/baselines") if f.startswith("metrics")]
+    files = [f for f in api.list_repo_files("illuin-cde/baselines-v2") if f.startswith("metrics")]
     print(files)
+
     metrics = []
+    cols = ["model", "is_contextual"]
     for file in files:
-        result_path = api.hf_hub_download("illuin-cde/baselines", filename=file)
+        result_path = api.hf_hub_download("illuin-cde/baselines-v2", filename=file)
         with open(result_path, "r") as f:
             dic = json.load(f)
-            dic.update(dic["metrics"])
+            metrics_cur = dic["metrics"]
+            for k, v in metrics_cur.items():
+                dic.update({k: v["ndcg_at_5"]})
+                cols.append(k)
             del dic["metrics"]
             metrics.append(dic)
 
     df = pd.DataFrame(metrics)
-    df = df[["model", "dataset", "split", "is_contextual", "ndcg_at_1", "ndcg_at_5", "ndcg_at_10", "ndcg_at_100"]]
+    df = df[cols]
     df["model"] = df["model"].apply(lambda x: x.split("/")[-1])
-    df["dataset"] = df["dataset"].apply(lambda x: x.split("/")[-1])
     # round all numeric columns
+    # avg all numeric columns
+    df["avg"] = df.iloc[:, 2:].mean(axis=1)
     df = df.round(3)
 
     # sort by ndcg_at_5
-    df = df.sort_values("ndcg_at_5", ascending=False)
+    df = df.sort_values(by="avg", ascending=False)
 
     # gradio display
     gradio_df = gr.Dataframe(df)