Perform chat completion inference :return: a tuple of assistant message dict, function calls dict, usage dict, and completion data dict
(self)
| 375 | yield tool_action_result_log_dict |
| 376 | |
| 377 | async def inference(self) -> Tuple[Dict, List, Dict, Dict]: |
| 378 | """ |
| 379 | Perform chat completion inference |
| 380 | :return: a tuple of assistant message dict, function calls dict, usage dict, and completion data dict |
| 381 | """ |
| 382 | |
| 383 | completion_data = await chat_completion( |
| 384 | model=self.model, |
| 385 | messages=self.chat_completion_messages, |
| 386 | functions=self.chat_completion_functions, |
| 387 | configs={}, |
| 388 | ) |
| 389 | |
| 390 | assistant_message_dict = completion_data["message"] |
| 391 | function_calls = assistant_message_dict.get("function_calls") |
| 392 | usage = completion_data.get("usage") |
| 393 | self.total_input_tokens += usage.get("input_tokens", 0) |
| 394 | self.total_output_tokens += usage.get("output_tokens", 0) |
| 395 | return assistant_message_dict, function_calls, usage, completion_data |
| 396 | |
| 397 | async def stream_inference(self, message_chunk_object_name="MessageChunk"): |
| 398 | try: |
no test coverage detected