reduce logs

src/benchmark.py

@@ -35,11 +35,11 @@ def make_request(url, headers, payload):
     response = requests.post(url, headers=headers, json=payload)
     return response
 
-url = "https://api.runpod.ai/v2/4hlrhh430u5tz7/run"
+url = "https://api.runpod.ai/v2/4hlrhh430u5tz7/status/1a73bdc5-aede-4c0a-8fa2-2ac2fb3010dc"
 
 while True:
     # Number of concurrent requests to make
-    num_requests = 100
+    num_requests = 1
 
     with concurrent.futures.ThreadPoolExecutor(max_workers=num_requests) as executor:
         futures = [executor.submit(make_request, url, headers, payload) for _ in range(num_requests)]
@@ -48,6 +48,7 @@ while True:
     for future in concurrent.futures.as_completed(futures):
         response = future.result()
         # Handle response as needed
+        print(response.json())
         print(response.status_code)
 
     # Sleep for 1 second before starting the next iteration

src/handler.py

@@ -15,6 +15,7 @@ MODEL_NAME = os.environ.get('MODEL_NAME')
 MODEL_BASE_PATH = os.environ.get('MODEL_BASE_PATH', '/runpod-volume/')
 STREAMING = os.environ.get('STREAMING', False) == 'True'
 TOKENIZER = os.environ.get('TOKENIZER', None)
+USE_FULL_METRICS = os.environ.get('USE_FULL_METRICS', False)
 
 if not MODEL_NAME:
     print("Error: The model has not been provided.")
@@ -173,7 +174,7 @@ async def handler_streaming(job: dict) -> Generator[dict[str, list], None, None]
                 text_outputs.append((" " if text_pos > 0 else "") + text_chunk)
 
         # Metrics for the vLLM serverless worker
-        runpod_metrics = prepare_metrics()
+        runpod_metrics = prepare_metrics() if USE_FULL_METRICS else {}
 
         # The input job
         runpod_metrics['job_input'] = job_input
@@ -257,7 +258,7 @@ async def handler(job: dict) -> dict[str, list]:
     num_seqs = sampling_params.n
 
     # Metrics for the vLLM serverless worker
-    runpod_metrics = prepare_metrics()
+    runpod_metrics = prepare_metrics() if USE_FULL_METRICS else {}
 
     # The input job
     runpod_metrics['job_input'] = job_input