Increase max_tokens for Gemini from 150 to 500 to prevent cut-off sentences

ml-service/app/models/llm_detoxifier.py

@@ -292,7 +292,7 @@ class LLMDetoxifier:
                     model=settings.GEMINI_MODEL,
                     messages=[{"role": "user", "content": prompt}],
                     temperature=0.3,
-                    max_tokens=150,
+                    max_tokens=500,
                 )
                 result = response.choices[0].message.content.strip() if response.choices and response.choices[0].message else None
             else:
@@ -334,7 +334,7 @@ class LLMDetoxifier:
                         model=settings.GEMINI_MODEL,
                         messages=[{"role": "user", "content": prompt}],
                         temperature=0.3,
-                        max_tokens=150,
+                        max_tokens=500,
                         stream=True,
                     )
                     full_generated = ""