misc: parse bench_serving result as markdown table (#9377)

test/srt/parse_results.py

@@ -8,6 +8,11 @@ from tabulate import tabulate
 # Parse command-line arguments
 parser = argparse.ArgumentParser(description="Parse JSONL benchmark and summarize.")
 parser.add_argument("input_file", type=str, help="Path to input JSONL file")
+parser.add_argument(
+    "--md",
+    action="store_true",
+    help="If set, print the summary table in Markdown format (GitHub style)",
+)
 args = parser.parse_args()
 
 input_file = args.input_file
@@ -44,5 +49,9 @@ df = pd.DataFrame(results)
 df.to_csv(output_file, index=False)
 print(f"\nSaved summary to: {output_file}\n")
 
-# Print ASCII table
-print(tabulate(df, headers="keys", tablefmt="grid", floatfmt=".3f"))
+if args.md:
+    # Print Markdown table
+    print(tabulate(df, headers="keys", tablefmt="github", floatfmt=".3f"))
+else:
+    # Print ASCII table
+    print(tabulate(df, headers="keys", tablefmt="grid", floatfmt=".3f"))