[BugFix] rollback max_tokens and min_tokens when continue to infer (#5082)

* [BugFix] rollback  max_tokens and min_tokens when continue to infer

* [BugFix] rollback  max_tokens and min_tokens when continue to infer

* [fix] add more logger info:  max_tokens

---------

Co-authored-by: liqinrui <liqinrui@baidu.com>

fastdeploy/entrypoints/engine_client.py

@@ -298,7 +298,12 @@ class EngineClient:
 
         if data.get("max_tokens") is not None:
             if data["max_tokens"] < 1 or data["max_tokens"] >= self.max_model_len:
-                raise ParameterError("max_tokens", f"max_tokens can be defined [1, {self.max_model_len}).")
+                api_server_logger.error(
+                    f"req_id:{data['request_id']}, max_tokens must be defined [1, {self.max_model_len}), but now it's {data['max_tokens']}."
+                )
+                raise ValueError(
+                    f"max_tokens can be defined [1, {self.max_model_len}), but now it's {data['max_tokens']}."
+                )
 
         if data.get("reasoning_max_tokens") is not None:
             if data["reasoning_max_tokens"] < 1:
@@ -306,7 +311,7 @@ class EngineClient:
             if data["reasoning_max_tokens"] > data["max_tokens"]:
                 data["reasoning_max_tokens"] = data["max_tokens"]
                 api_server_logger.warning(
-                    f"req_id: {data['request_id']}, reasoning_max_tokens exceeds max_tokens, the value of reasoning_max_tokens will be adjusted to match that of max_tokens"
+                    f"req_id: {data['request_id']}, reasoning_max_tokens exceeds max_tokens, the value of reasoning_max_tokens will be adjusted to {data['max_tokens']}"
                 )
         if data.get("temperature") is not None and abs(data["temperature"]) < 1e-6:
             data["temperature"] = 1e-6

fastdeploy/input/ernie4_5_vl_processor/ernie4_5_vl_processor.py

@@ -264,14 +264,12 @@ class Ernie4_5_VLProcessor(Ernie4_5Processor):
         if max_model_len is not None and len(request["prompt_token_ids"]) > max_model_len:
             request["prompt_token_ids"] = request["prompt_token_ids"][: max_model_len - 1]
 
-        tmp_max_tokens = 0
         if request.get("max_tokens") is None:
             request["max_tokens"] = max(1, max_model_len - len(request["prompt_token_ids"]))
-            tmp_max_tokens = request["max_tokens"]
         else:
             request["max_tokens"] = min(max_model_len - len(request["prompt_token_ids"]), request["max_tokens"])
         if request.get("reasoning_max_tokens") is None:
-            request["reasoning_max_tokens"] = max(int(tmp_max_tokens * 0.8), 1)
+            request["reasoning_max_tokens"] = max(int(request["max_tokens"] * 0.8), 1)
         data_processor_logger.info(f"Processed request {request}")
 
         if request.get("top_p") is not None and request.get("top_p") < _SAMPLING_EPS: