get default system_prompt strings on each generate call

ragna/assistants/_ai21labs.py

@@ -44,8 +44,8 @@ def _render_prompt(self, prompt: Union[str, list[Message]]) -> Union[str, list]:
     async def generate(
         self,
         prompt: Union[str, list[Message]],
-        system_prompt: str,
         *,
+        system_prompt: str = "You are a helpful assistant.",
         max_new_tokens: int = 256,
     ) -> AsyncIterator[str]:
         """

ragna/assistants/_anthropic.py

@@ -58,8 +58,8 @@ def _render_prompt(self, prompt: Union[str, list[Message]]) -> list[dict]:
     async def generate(
         self,
         prompt: Union[str, list[Message]],
-        system_prompt: str,
         *,
+        system_prompt: str = "You are a helpful assistant.",
         max_new_tokens: int = 256,
     ) -> AsyncIterator[str]:
         """

ragna/assistants/_cohere.py

@@ -42,9 +42,9 @@ def _render_prompt(self, prompt: Union[str, list[Message]]) -> str:
     async def generate(
         self,
         prompt: Union[str, list[Message]],
-        system_prompt: str,
         source_documents: list[dict[str, str]],
         *,
+        system_prompt: str = "You are a helpful assistant.",
         max_new_tokens: int = 256,
     ) -> AsyncIterator[str]:
         """

ragna/assistants/_openai.py

@@ -39,7 +39,7 @@ def _render_prompt(self, prompt: Union[str,list[Message]], system_prompt: str) -
         return system_message.extend(messages)
 
     async def generate(
-        self, prompt: Union[str,list[Message]], system_prompt: str, *, max_new_tokens: int = 256
+        self, prompt: Union[str,list[Message]], *, system_prompt: str = "You are a helpful assistant.", max_new_tokens: int = 256
     ) -> AsyncIterator[dict[str, Any]]:
             """
         Primary method for calling assistant inference, either as a one-off request from anywhere in ragna, or as part of self.answer()