fix bug

app.py

@@ -20,7 +20,7 @@ COLUMNS_MAPPING = {
     "num_parameters": "#Parameters 📏",
     "forward.peak_memory(MB)": "Peak Memory (MB) ⬇️",
     "generate.throughput(tokens/s)": "Throughput (tokens/s) ⬆️",
-    "h4_score": "Average Open LLM Score ⬆️",
+    "average": "Average Open LLM Score ⬆️",
 }
 COLUMNS_DATATYPES = ["markdown", "str", "str",
                      "number", "number", "number", "markdown"]
@@ -67,7 +67,7 @@ def get_benchmark_plot(bench_df):
     bench_df = bench_df[bench_df["generate.latency(s)"] < 100]
 
     fig = px.scatter(
-        bench_df, x="generate.latency(s)", y="h4_score",
+        bench_df, x="generate.latency(s)", y="average",
         color='model_type', symbol='backend.name', size='forward.peak_memory(MB)',
         custom_data=['model', 'backend.name', 'backend.torch_dtype',
                      'forward.peak_memory(MB)', 'generate.throughput(tokens/s)'],
@@ -120,7 +120,7 @@ def filter_query(text, backends, datatypes, threshold, benchmark="1xA100-80GB"):
         raw_df["model"].str.lower().str.contains(text.lower()) &
         raw_df["backend.name"].isin(backends) &
         raw_df["backend.torch_dtype"].isin(datatypes) &
-        (raw_df["h4_score"] >= threshold)
+        (raw_df["average"] >= threshold)
     ]
 
     filtered_table = get_benchmark_table(filtered_df)