Merge pull request #2 from vllm-project/fix-remove-no-need

fix: remove no needed todo and verify CI

Author: rootfs

bench/bench_plot.py

@@ -178,11 +178,16 @@ def plot_metric(metric: str, out_path: Path):
     # Determine modes to plot, optionally limiting to top-N by mean of metric
     all_modes = sorted({m for c in cats for m in cat_by_mode.get(c, {}).keys()})
     if len(all_modes) > 0:
+
         def _mean(values):
             vals = [v for v in values if v is not None]
             return sum(vals) / len(vals) if vals else float("nan")
 
-        if args.max_modes is not None and args.max_modes > 0 and len(all_modes) > args.max_modes:
+        if (
+            args.max_modes is not None
+            and args.max_modes > 0
+            and len(all_modes) > args.max_modes
+        ):
             mode_means = []
             for mode in all_modes:
                 vals = [cat_by_mode.get(c, {}).get(mode, {}).get(metric) for c in cats]
@@ -279,7 +284,9 @@ def _mean(values):
             ax.set_xlim(left_xlim, right_xlim)
     ylabel = metric.replace("_", " ")
     ax.set_ylabel(ylabel, fontsize=int(18 * args.font_scale))
-    ax.set_title(f"Per-category {ylabel} per-mode values", fontsize=int(22 * args.font_scale))
+    ax.set_title(
+        f"Per-category {ylabel} per-mode values", fontsize=int(22 * args.font_scale)
+    )
     ax.tick_params(axis="both", which="major", labelsize=int(14 * args.font_scale))
 
     # Build a figure-level legend below the axes and reserve space to prevent overlap
@@ -290,7 +297,7 @@ def _mean(values):
         legend_rows = 2
         legend_ncol = max(1, (num_series + legend_rows - 1) // legend_rows)
         num_rows = legend_rows
-        scale = (args.font_scale / 1.6)
+        scale = args.font_scale / 1.6
         # Reserve generous space for long rotated tick labels and multi-row legend
         bottom_reserved = (0.28 * scale) + (0.12 * num_rows * scale)
         bottom_reserved = max(0.24, min(0.60, bottom_reserved))

bench/router_reason_bench.py

@@ -355,7 +355,9 @@ def build_extra_body_for_model(
     # reasoning: True -> ON, False -> OFF, None -> base
 
     lower = model_name.lower()
-    if (("ds" in lower) or ("deepseek" in lower)) and ("v31" in lower or "v3.1" in lower or "v3" in lower):
+    if (("ds" in lower) or ("deepseek" in lower)) and (
+        "v31" in lower or "v3.1" in lower or "v3" in lower
+    ):
         if reasoning is True:
             return {"chat_template_kwargs": {"thinking": True}}
         if reasoning is None or reasoning is False:
@@ -535,7 +537,9 @@ def run_variants(q: Dict[str, Any]) -> List[Dict[str, Any]]:
 
     with ThreadPoolExecutor(max_workers=concurrent_requests) as executor:
         futures = [executor.submit(run_variants, q) for q in questions_data]
-        for future in tqdm(futures, total=len(futures), desc=f"Evaluating {model} (vLLM modes)"):
+        for future in tqdm(
+            futures, total=len(futures), desc=f"Evaluating {model} (vLLM modes)"
+        ):
             results.extend(future.result())
 
     return pd.DataFrame(results)
@@ -584,7 +588,9 @@ def run_all_modes(q: Dict[str, Any]) -> List[Dict[str, Any]]:
 
     with ThreadPoolExecutor(max_workers=concurrent_requests) as executor:
         futures = [executor.submit(run_all_modes, q) for q in questions]
-        for future in tqdm(futures, total=len(futures), desc=f"Evaluating {model} (policies)"):
+        for future in tqdm(
+            futures, total=len(futures), desc=f"Evaluating {model} (policies)"
+        ):
             per_call_records.extend(future.result())
 
     calls_df = pd.DataFrame(per_call_records)

src/semantic-router/pkg/extproc/request_handler.go

@@ -445,7 +445,6 @@ func (r *OpenAIRouter) handleModelRouting(openAIRequest *openai.ChatCompletionNe
 			selectedEndpoint = endpointAddress
 			log.Printf("Selected endpoint address: %s for model: %s", selectedEndpoint, originalModel)
 		} else {
-			// TOOD(Xunzhuo): pick a random endpoint from the list of all available endpoints
 			log.Printf("Warning: No endpoint found for model %s, using fallback", originalModel)
 		}
 		setHeaders := []*core.HeaderValueOption{}