Addressing hwoo's CR

model_analyzer/config/generate/generator_utils.py

@@ -118,11 +118,13 @@ def generate_doubled_list(min_value: int, max_value: int) -> List[int]:
         return list
 
     @staticmethod
-    def extract_max_tokens_from_request_parameter(request_parameter: str) -> int:
-        # format is max_tokens:<num>:int
-        _, max_tokens, _ = request_parameter.split(":")
+    def extract_value_from_request_parameter(request_parameter: str) -> int:
+        # Format is: <parameter>:<value>:<type>
+        # Example: max_tokens:10:int
 
-        return int(max_tokens)
+        _, value, _ = request_parameter.split(":")
+
+        return int(value)
 
     @staticmethod
     def extract_text_input_length_from_input_data(input_data: str) -> int:

model_analyzer/config/generate/perf_analyzer_config_generator.py

@@ -395,9 +395,7 @@ def _update_perf_config_based_on_parameter_combination(
     ) -> None:
         if "request-parameter" in parameter_combination:
             request_parameter = parameter_combination["request-parameter"]
-            max_tokens = utils.extract_max_tokens_from_request_parameter(
-                request_parameter
-            )
+            max_tokens = utils.extract_value_from_request_parameter(request_parameter)
             parameter_combination["request-period"] = (
                 max_tokens
                 if max_tokens < parameter_combination["request-period"]

model_analyzer/perf_analyzer/perf_config.py

@@ -284,7 +284,7 @@ def extract_model_specific_parameters(self):
             "concurrency-range": self._args["concurrency-range"],
             "request-rate-range": self._args["request-rate-range"],
             "periodic-concurrency-range": self._args["periodic-concurrency-range"],
-            "max-tokens": utils.extract_max_tokens_from_request_parameter(
+            "max-tokens": utils.extract_value_from_request_parameter(
                 self._args["request-parameter"]
             ),
             "request-period": self._args["request-period"],

model_analyzer/record/types/avg_first_token_latency.py

@@ -57,7 +57,7 @@ def header(cls, aggregation_tag=False):
             metric.
         """
 
-        return "Avg First Token-to-Token latency (ms)"
+        return "Avg First Token latency (ms)"
 
     def __eq__(self, other):
         """

model_analyzer/result/result_table_manager.py

@@ -39,7 +39,7 @@ class ResultTableManager:
         "concurrency": "Concurrency",
         "periodic_concurrency": "Periodic Concurrency",
         "text_input_length": "Text Input Length",
-        "max_tokens": "Max Token Count",
+        "max_tokens": "Max Tokens",
         "request_period": "Request Period",
         "request_rate": "Request Rate",
         "model_config_path": "Model Config Path",