updated

Author: kenoharada

README.md

@@ -17,7 +17,6 @@ from llm_api_utils import get_llm_response
 
 model_name = 'gpt-4o-mini-2024-07-18'
 params = {
-    'max_tokens': 256, 
     'temperature': 0.0
 }
 
@@ -56,7 +55,7 @@ print('#######', model_name)
 [Sample code](./call_gpt.py)
 - document: https://platform.openai.com/docs/overview  
 - library: https://github.com/openai/openai-python  
-- models: https://platform.openai.com/docs/models  
+- models: https://platform.openaxi.com/docs/models  
 - playground: https://platform.openai.com/playground/chat?models=gpt-4o  
 - pricing: https://openai.com/api/pricing/  
 - status: https://status.openai.com/  

example.py

@@ -3,7 +3,6 @@
 
 model_name = 'gpt-4o-mini-2024-07-18'
 params = {
-    'max_tokens': 4096, 
     'temperature': 0.0
 }
 

llm_api_utils.py

@@ -13,14 +13,17 @@
 import google.api_core.exceptions as google_exceptions
 
 
-def get_llm_response(model_name: str, params: dict, messages: list[dict]) -> str:
+def get_llm_response(model_name: str, params_: dict, messages: list[dict]) -> str:
+    params = params_.copy()
     if model_name in OPENAI_MODEL_NAMES:
         if 'max_tokens' in params:
-            params_copy = params.copy()
-            params_copy['max_completion_tokens'] = params_copy['max_tokens']
-            del params_copy['max_tokens']
-        return get_gpt_respnose(model_name, params_copy, messages)
+            params['maxcompletion_tokens'] = params['max_tokens']
+            del params['max_tokens']
+            return get_gpt_respnose(model_name, params, messages)
+        return get_gpt_respnose(model_name, params, messages)
     elif model_name in ANTHROPIC_MODEL_NAMES:
+        if 'max_tokens' not in params:
+            params['max_tokens'] = 8192
         return get_claude_response(model_name, params, messages)
     elif model_name in GEMINI_MODEL_NAMES:
         return get_gemini_response(model_name, params, messages)
@@ -32,9 +35,11 @@ def get_llm_response(model_name: str, params: dict, messages: list[dict]) -> str
 async def get_llm_response_async(model_name: str, params: dict, messages: list[dict]) -> str:
     if model_name in OPENAI_MODEL_NAMES:
         if 'max_tokens' in params:
-            params['max_completion_tokens'] = params['max_tokens']
-            del params['max_tokens']
-        return await get_gpt_respnose_async(model_name, params, messages)
+            params_copy = params.copy()
+            params_copy['max_completion_tokens'] = params_copy['max_tokens']
+            del params_copy['max_tokens']
+            return get_gpt_respnose(model_name, params_copy, messages)
+        return get_gpt_respnose(model_name, params, messages)
     elif model_name in ANTHROPIC_MODEL_NAMES:
         return await get_claude_response_async(model_name, params, messages)
     elif model_name in GEMINI_MODEL_NAMES: