""" # Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License" # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """ import asyncio import itertools import time import traceback import uuid from collections.abc import Iterable from typing import List, Optional import numpy as np from fastdeploy.entrypoints.openai.protocol import ( ChatCompletionRequest, ChatCompletionResponse, ChatCompletionResponseChoice, ChatCompletionResponseStreamChoice, ChatCompletionStreamResponse, ChatMessage, CompletionTokenUsageInfo, DeltaMessage, ErrorInfo, ErrorResponse, LogProbEntry, LogProbs, PromptTokenUsageInfo, UsageInfo, ) from fastdeploy.entrypoints.openai.response_processors import ChatResponseProcessor from fastdeploy.metrics.metrics import main_process_metrics from fastdeploy.trace.constants import LoggingEventName from fastdeploy.trace.trace_logger import print as trace_print from fastdeploy.utils import ( ErrorCode, ErrorType, ParameterError, api_server_logger, clamp_prompt_logprobs, get_host_ip, ) from fastdeploy.worker.output import ( Logprob, LogprobsLists, LogprobsTensors, PromptLogprobs, SpeculateMetrics, ) NONES = itertools.repeat(None) class OpenAIServingChat: """ OpenAI-style chat completions serving """ def __init__( self, engine_client, models, pid, ips, max_waiting_time, chat_template, enable_mm_output: Optional[bool] = False, tokenizer_base_url: Optional[str] = None, ): self.engine_client = engine_client self.models = models self.pid = pid self.max_waiting_time = max_waiting_time self.chat_template = chat_template self.enable_mm_output = enable_mm_output self.tokenizer_base_url = tokenizer_base_url if ips is not None: if isinstance(ips, list): self.master_ip = ips[0] else: self.master_ip = ips.split(",")[0] self.is_master_ip = get_host_ip() == self.master_ip else: self.master_ip = "0.0.0.0" self.is_master_ip = True api_server_logger.info(f"master ip: {self.master_ip}") def _check_master(self): return self.engine_client.is_master or self.is_master_ip async def create_chat_completion(self, request: ChatCompletionRequest): """ Create a new chat completion using the specified parameters. """ if not self._check_master(): err_msg = ( f"Only master node can accept completion request, please send request to master node: {self.master_ip}" ) api_server_logger.error(err_msg) return ErrorResponse(error=ErrorInfo(message=err_msg, type=ErrorType.INTERNAL_ERROR)) if self.models: is_supported, request.model = self.models.is_supported_model(request.model) if not is_supported: err_msg = f"Unsupported model: [{request.model}], support [{', '.join([x.name for x in self.models.model_paths])}] or default" api_server_logger.error(err_msg) return ErrorResponse( error=ErrorInfo(message=err_msg, type=ErrorType.INTERNAL_ERROR, code=ErrorCode.MODEL_NOT_SUPPORT) ) try: if self.max_waiting_time < 0: await self.engine_client.semaphore.acquire() else: await asyncio.wait_for(self.engine_client.semaphore.acquire(), timeout=self.max_waiting_time) api_server_logger.info(f"current {self.engine_client.semaphore.status()}") if request.request_id is not None: request_id = request.request_id if not request_id.startswith("chatcmpl-"): request_id = f"chatcmpl-{request_id}" elif request.user is not None: request_id = f"chatcmpl-{request.user}-{uuid.uuid4()}" else: request_id = f"chatcmpl-{uuid.uuid4()}" api_server_logger.info(f"create chat completion request: {request_id}") prompt_tokens = None max_tokens = None try: current_req_dict = request.to_dict_for_infer(f"{request_id}_0") if "chat_template" not in current_req_dict: current_req_dict["chat_template"] = self.chat_template current_req_dict["arrival_time"] = time.time() # preprocess the req_dict prompt_token_ids = await self.engine_client.format_and_add_data(current_req_dict) prompt_tokens = current_req_dict.get("prompt_tokens") max_tokens = current_req_dict.get("max_tokens") if isinstance(prompt_token_ids, np.ndarray): prompt_token_ids = prompt_token_ids.tolist() except ParameterError as e: api_server_logger.error(f"request[{request_id}] generator error: {str(e)}, {e.message}") self.engine_client.semaphore.release() return ErrorResponse( error=ErrorInfo(message=str(e.message), type=ErrorType.INVALID_REQUEST_ERROR, param=e.param) ) except Exception as e: error_msg = f"request[{request_id}] generator error: {str(e)}, {str(traceback.format_exc())}" api_server_logger.error(error_msg) self.engine_client.semaphore.release() return ErrorResponse(error=ErrorInfo(message=error_msg, type=ErrorType.INVALID_REQUEST_ERROR)) del current_req_dict if request.stream: return self.chat_completion_stream_generator( request, request_id, request.model, prompt_token_ids, prompt_tokens, max_tokens ) else: try: return await self.chat_completion_full_generator( request, request_id, request.model, prompt_token_ids, prompt_tokens, max_tokens ) except Exception as e: error_msg = f"request[{request_id}]full generator error: {str(e)}, {str(traceback.format_exc())}" api_server_logger.error(error_msg) return ErrorResponse(error=ErrorInfo(message=error_msg, type=ErrorType.INTERNAL_ERROR)) except Exception as e: error_msg = ( f"request[{request_id}] waiting error: {str(e)}, {str(traceback.format_exc())}, " f"max waiting time: {self.max_waiting_time}" ) api_server_logger.error(error_msg) return ErrorResponse( error=ErrorInfo(message=error_msg, type=ErrorType.TIMEOUT_ERROR, code=ErrorCode.TIMEOUT) ) def _create_streaming_error_response(self, message: str) -> str: api_server_logger.error(message) error_response = ErrorResponse(error=ErrorInfo(message=message, type=ErrorType.INTERNAL_ERROR)) return error_response.model_dump_json() async def chat_completion_stream_generator( self, request: ChatCompletionRequest, request_id: str, model_name: str, prompt_token_ids: list(), prompt_tokens: str, max_tokens: int, ): """ Streaming chat completion generator. """ created_time = int(time.time()) chunk_object_type: str = "chat.completion.chunk" num_choices = 1 if request.n is None else request.n first_iteration = True previous_num_tokens = [0] * num_choices reasoning_num_tokens = [0] * num_choices num_prompt_tokens = 0 num_cached_tokens = 0 num_image_tokens = [0] * num_choices tool_called = [False] * num_choices inference_start_time = [0] * num_choices max_streaming_response_tokens = ( request.max_streaming_response_tokens if request.max_streaming_response_tokens is not None else (request.metadata or {}).get("max_streaming_response_tokens", 1) ) # dierctly passed & passed in metadata max_streaming_response_tokens = max(1, max_streaming_response_tokens) enable_thinking = self._get_thinking_status(request) include_stop_str_in_output = request.include_stop_str_in_output stream_options = request.stream_options if stream_options is None: include_usage = False include_continuous_usage = False else: include_usage = stream_options.include_usage include_continuous_usage = stream_options.continuous_usage_stats chunk = ChatCompletionStreamResponse( id=request_id, object=chunk_object_type, created=created_time, choices=[], model=model_name, ) api_server_logger.info(f"create chat completion request: {request_id}") try: dealer, response_queue = await self.engine_client.connection_manager.get_connection( request_id, num_choices ) request_ids = [f"{request_id}_{i}" for i in range(num_choices)] for rid in request_ids: dealer.write([b"", rid.encode("utf-8")]) choices = [] current_waiting_time = 0 response_processor = ChatResponseProcessor( data_processor=self.engine_client.data_processor, enable_mm_output=self.enable_mm_output, decoder_base_url=self.tokenizer_base_url, ) while num_choices > 0: if self.engine_client.check_model_weight_status(): raise ValueError("Engine is clearing model weight") try: response = await asyncio.wait_for(response_queue.get(), timeout=10) current_waiting_time = 0 except asyncio.TimeoutError: current_waiting_time += 10 if current_waiting_time == 300: status, msg = self.engine_client.check_health() if not status: if choices: chunk.choices = choices yield f"data: {chunk.model_dump_json(exclude_unset=True)}\n\n" raise ValueError(f"Engine is not healthy: {msg}") else: current_waiting_time = 0 await asyncio.sleep(0.01) continue generator = response_processor.process_response_chat( response, stream=True, enable_thinking=enable_thinking, include_stop_str_in_output=include_stop_str_in_output, ) async for res in generator: idx = int(res["request_id"].split("_")[-1]) if res.get("error_code", 200) != 200: raise ValueError("{}".format(res["error_msg"])) if res["metrics"]["first_token_time"] is not None: arrival_time = res["metrics"]["first_token_time"] inference_start_time[idx] = res["metrics"]["inference_start_time"] else: arrival_time = res["metrics"]["arrival_time"] - inference_start_time[idx] if first_iteration: num_prompt_tokens = len(prompt_token_ids) num_cached_tokens = res.get("num_cached_tokens", 0) num_input_image_tokens = res.get("num_input_image_tokens", 0) num_input_video_tokens = res.get("num_input_video_tokens", 0) for i in range(num_choices): prompt_logprobs_res: Optional[PromptLogprobs] = None prompt_logprobs_tensors = res.get("prompt_logprobs", None) if request.prompt_logprobs is not None and prompt_logprobs_tensors is not None: num_prompt_logprobs = ( request.prompt_logprobs if request.prompt_logprobs != -1 else self.engine_client.ori_vocab_size ) prompt_logprobs_res = self._build_prompt_logprobs( prompt_logprobs_tensors, num_prompt_logprobs, request.include_logprobs_decode_token ) choice = ChatCompletionResponseStreamChoice( index=i, delta=DeltaMessage( role="assistant", reasoning_content="", tool_calls=None, prompt_token_ids=None, completion_token_ids=None, ), prompt_logprobs=clamp_prompt_logprobs(prompt_logprobs_res), ) if response_processor.enable_multimodal_content(): choice.delta.multimodal_content = [ { "type": "text", "text": "", } ] else: choice.delta.content = "" if res["outputs"].get("audio_content", None) is not None: choice.delta.audio_content = res["outputs"]["audio_content"] if request.return_token_ids: choice.delta.prompt_token_ids = list(prompt_token_ids) choice.delta.prompt_tokens = prompt_tokens chunk = ChatCompletionStreamResponse( id=request_id, object=chunk_object_type, created=created_time, choices=[choice], model=model_name, ) if include_continuous_usage: chunk.usage = UsageInfo( prompt_tokens=num_prompt_tokens, completion_tokens=0, total_tokens=num_prompt_tokens, prompt_tokens_details=PromptTokenUsageInfo( cached_tokens=num_cached_tokens, image_tokens=num_input_image_tokens, video_tokens=num_input_video_tokens, ), completion_tokens_details=CompletionTokenUsageInfo(reasoning_tokens=0), ) yield f"data: {chunk.model_dump_json(exclude_unset=True)} \n\n" api_server_logger.info(f"Chat Streaming response send_idx 0: {chunk.model_dump_json()}") first_iteration = False output = res["outputs"] output_top_logprobs = output["top_logprobs"] output_draft_top_logprobs = output["draft_top_logprobs"] previous_num_tokens[idx] += len(output["token_ids"]) if output.get("num_image_tokens"): previous_num_tokens[idx] += output.get("num_image_tokens") num_image_tokens[idx] += output.get("num_image_tokens") reasoning_num_tokens[idx] += output.get("reasoning_token_num", 0) logprobs_res: Optional[LogProbs] = None draft_logprobs_res: Optional[LogProbs] = None if request.logprobs and output_top_logprobs is not None: num_top_logprobs = ( request.top_logprobs if request.top_logprobs != -1 else self.engine_client.ori_vocab_size ) logprobs_res = self._create_chat_logprobs( output_top_logprobs, request.logprobs, num_top_logprobs, request.include_logprobs_decode_token, ) if request.include_draft_logprobs and output_draft_top_logprobs is not None: draft_logprobs_res = self._create_chat_logprobs( output_draft_top_logprobs, request.logprobs, num_top_logprobs, request.include_logprobs_decode_token, ) output_speculate_metrics = res["metrics"].get("speculate_metrics", None) delta_message = DeltaMessage( reasoning_content="", prompt_token_ids=None, tool_calls=None, completion_token_ids=None, ) if response_processor.enable_multimodal_content(): delta_message.multimodal_content = output["multipart"] else: delta_message.content = output["text"] if output.get("audio_content", None) is not None: delta_message.audio_content = output["audio_content"] if not res["finished"] and "delta_message" in output: delta_message_output = output["delta_message"] if delta_message_output is None: continue delta_message.content = delta_message_output.content or "" delta_message.reasoning_content = delta_message_output.reasoning_content or "" if delta_message_output.tool_calls: delta_message.tool_calls = delta_message_output.tool_calls tool_called[idx] = True choice = ChatCompletionResponseStreamChoice( index=idx, delta=delta_message, logprobs=logprobs_res, draft_logprobs=draft_logprobs_res, arrival_time=arrival_time, speculate_metrics=output_speculate_metrics, ) if res["finished"]: num_choices -= 1 main_process_metrics.e2e_request_latency.observe( time.time() - res["metrics"]["request_start_time"] ) if previous_num_tokens[idx] != max_tokens: choice.finish_reason = "stop" if tool_called[idx]: choice.finish_reason = "tool_calls" else: choice.finish_reason = "length" if res.get("error_msg") is not None and "Recover" in res["error_msg"]: choice.finish_reason = "recover_stop" if request.return_token_ids: if response_processor.enable_multimodal_content(): choice.delta.multimodal_content[0]["completion_token_ids"] = list(output["token_ids"]) else: choice.delta.completion_token_ids = list(output["token_ids"]) choice.delta.completion_tokens = output.get("completion_tokens") if include_continuous_usage: chunk.usage = UsageInfo( prompt_tokens=num_prompt_tokens, completion_tokens=previous_num_tokens[idx], total_tokens=num_prompt_tokens + previous_num_tokens[idx], prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=num_cached_tokens), completion_tokens_details=CompletionTokenUsageInfo( reasoning_tokens=reasoning_num_tokens[idx], image_tokens=num_image_tokens[idx], ), ) choices.append(choice) if len(choices) == max_streaming_response_tokens or res["finished"]: chunk.choices = choices yield f"data: {chunk.model_dump_json(exclude_unset=True)}\n\n" if res["finished"]: api_server_logger.info(f"Chat Streaming response last send: {chunk.model_dump_json()}") choices = [] if include_usage: completion_tokens = sum(previous_num_tokens) reasoning_tokens = sum(reasoning_num_tokens) usage = UsageInfo( prompt_tokens=num_prompt_tokens, completion_tokens=completion_tokens, total_tokens=num_prompt_tokens + completion_tokens, prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=num_cached_tokens), completion_tokens_details=CompletionTokenUsageInfo( image_tokens=sum(num_image_tokens), reasoning_tokens=reasoning_tokens ), ) chunk = ChatCompletionStreamResponse( id=request_id, object=chunk_object_type, created=created_time, choices=[], model=model_name, usage=usage, ) yield f"data: {chunk.model_dump_json(exclude_unset=True)}\n\n" except Exception as e: error_data = self._create_streaming_error_response( f"request[{request_id}] generate stream error: {str(e)}, {str(traceback.format_exc())}" ) yield f"data: {error_data}\n\n" finally: await self.engine_client.connection_manager.cleanup_request(request_id) self.engine_client.semaphore.release() trace_print(LoggingEventName.POSTPROCESSING_END, request_id, getattr(request, "user", "")) api_server_logger.info(f"release {request_id} {self.engine_client.semaphore.status()}") yield "data: [DONE]\n\n" async def chat_completion_full_generator( self, request: ChatCompletionRequest, request_id: str, model_name: str, prompt_token_ids: list(), prompt_tokens: str, max_tokens: int, ): """ Full chat completion generator. """ created_time = int(time.time()) num_choices = 1 if request.n is None else request.n enable_thinking = self._get_thinking_status(request) include_stop_str_in_output = request.include_stop_str_in_output try: dealer, response_queue = await self.engine_client.connection_manager.get_connection( request_id, num_choices ) # dealer.write([b"", request_id.encode("utf-8")]) request_ids = [f"{request_id}_{i}" for i in range(num_choices)] for rid in request_ids: dealer.write([b"", rid.encode("utf-8")]) previous_num_tokens = [0] * num_choices reasoning_num_tokens = [0] * num_choices current_waiting_time = 0 logprob_contents = [[] for _ in range(num_choices)] draft_logprob_contents = [[] for _ in range(num_choices)] completion_token_ids = [[] for _ in range(num_choices)] num_cached_tokens = [0] * num_choices num_input_image_tokens = [0] * num_choices num_input_video_tokens = [0] * num_choices num_image_tokens = [0] * num_choices response_processor = ChatResponseProcessor( data_processor=self.engine_client.data_processor, enable_mm_output=self.enable_mm_output, decoder_base_url=self.tokenizer_base_url, ) prompt_logprobs_res_list = [[] for _ in range(num_choices)] speculate_metrics = [None for _ in range(num_choices)] choices = [] while num_choices > 0: if self.engine_client.check_model_weight_status(): return ErrorResponse( error=ErrorInfo( message="Model weight cleared", code=ErrorCode.INVALID_VALUE, type=ErrorType.INVALID_REQUEST_ERROR, ) ) try: response = await asyncio.wait_for(response_queue.get(), timeout=10) current_waiting_time = 0 except asyncio.TimeoutError: current_waiting_time += 10 if current_waiting_time == 300: status, msg = self.engine_client.check_health() if not status: raise ValueError(f"Engine is not healthy: {msg}") else: current_waiting_time = 0 await asyncio.sleep(0.1) continue generator = response_processor.process_response_chat( response, stream=False, enable_thinking=enable_thinking, include_stop_str_in_output=include_stop_str_in_output, ) async for data in generator: if data.get("error_code", 200) != 200: raise ValueError("{}".format(data["error_msg"])) idx = int(data["request_id"].split("_")[-1]) # api_server_logger.debug(f"Client {request_id} received: {data}") previous_num_tokens[idx] += len(data["outputs"]["token_ids"]) completion_token_ids[idx].extend(data["outputs"]["token_ids"]) # The logprob for handling the response output = data["outputs"] output_top_logprobs = output["top_logprobs"] output_draft_top_logprobs = output["draft_top_logprobs"] if output_top_logprobs is not None: num_top_logprobs = ( request.top_logprobs if request.top_logprobs != -1 else self.engine_client.ori_vocab_size ) # logprobs logprobs_res = self._create_chat_logprobs( output_top_logprobs, request.logprobs, num_top_logprobs, request.include_logprobs_decode_token, ) if logprobs_res and logprobs_res.content is not None: logprob_contents[idx].extend(logprobs_res.content) # draft_logprobs if request.include_draft_logprobs and output_draft_top_logprobs is not None: draft_logprobs_res = self._create_chat_logprobs( output_draft_top_logprobs, request.logprobs, num_top_logprobs, request.include_logprobs_decode_token, ) if draft_logprobs_res and draft_logprobs_res.content is not None: draft_logprob_contents[idx].extend(draft_logprobs_res.content) prompt_logprobs_tensors = data.get("prompt_logprobs", None) if request.prompt_logprobs is not None and prompt_logprobs_tensors is not None: num_prompt_logprobs = ( request.prompt_logprobs if request.prompt_logprobs != -1 else self.engine_client.ori_vocab_size ) prompt_logprobs_res = self._build_prompt_logprobs( prompt_logprobs_tensors, num_prompt_logprobs, request.include_logprobs_decode_token ) if prompt_logprobs_res: prompt_logprobs_res_list[idx].extend(clamp_prompt_logprobs(prompt_logprobs_res)) speculate_metrics[idx] = data["metrics"].get("speculate_metrics", None) if data["finished"]: num_choices -= 1 reasoning_num_tokens[idx] = data["outputs"].get("reasoning_token_num", 0) if data["outputs"].get("image_token_num"): previous_num_tokens[idx] += data["outputs"].get("image_token_num") num_image_tokens[idx] = data["outputs"].get("image_token_num") choice = await self._create_chat_completion_choice( data=data, request=request, prompt_token_ids=prompt_token_ids, prompt_tokens=prompt_tokens, completion_token_ids=completion_token_ids[idx], previous_num_tokens=previous_num_tokens[idx], num_cached_tokens=num_cached_tokens, num_input_image_tokens=num_input_image_tokens, num_input_video_tokens=num_input_video_tokens, num_image_tokens=num_image_tokens, logprob_contents=logprob_contents, draft_logprob_contents=draft_logprob_contents, response_processor=response_processor, prompt_logprobs_res_list=prompt_logprobs_res_list, max_tokens=max_tokens, speculate_metrics=speculate_metrics[idx], ) choices.append(choice) finally: await self.engine_client.connection_manager.cleanup_request(request_id) self.engine_client.semaphore.release() api_server_logger.info(f"release {self.engine_client.semaphore.status()}") num_prompt_tokens = len(prompt_token_ids) num_generated_tokens = sum(previous_num_tokens) num_reasoning_tokens = sum(reasoning_num_tokens) usage = UsageInfo( prompt_tokens=num_prompt_tokens, completion_tokens=num_generated_tokens, total_tokens=num_prompt_tokens + num_generated_tokens, prompt_tokens_details=PromptTokenUsageInfo( cached_tokens=sum(num_cached_tokens), image_tokens=sum(num_input_image_tokens), video_tokens=sum(num_input_video_tokens), ), completion_tokens_details=CompletionTokenUsageInfo( reasoning_tokens=num_reasoning_tokens, image_tokens=sum(num_image_tokens) ), ) choices = sorted(choices, key=lambda x: x.index) res = ChatCompletionResponse( id=request_id, created=created_time, model=model_name, choices=choices, usage=usage, ) trace_print(LoggingEventName.POSTPROCESSING_END, request_id, getattr(request, "user", "")) api_server_logger.info(f"Chat response: {res.model_dump_json()}") return res async def _create_chat_completion_choice( self, data: dict, request: ChatCompletionRequest, prompt_token_ids: list, prompt_tokens: str, completion_token_ids: list, previous_num_tokens: int, num_cached_tokens: list, num_input_image_tokens: list, num_input_video_tokens: list, num_image_tokens: list, logprob_contents: list, draft_logprob_contents: list, prompt_logprobs_res_list: list, response_processor: ChatResponseProcessor, max_tokens: int, speculate_metrics: SpeculateMetrics | None, ) -> ChatCompletionResponseChoice: idx = int(data["request_id"].split("_")[-1]) output = data["outputs"] if output is not None and output.get("metrics") and output["metrics"].get("request_start_time"): main_process_metrics.e2e_request_latency.observe( time.time() - data.get("metrics").get("request_start_time") ) message = ChatMessage( role="assistant", reasoning_content=output.get("reasoning_content"), tool_calls=output.get("tool_call"), prompt_token_ids=prompt_token_ids if request.return_token_ids else None, completion_token_ids=completion_token_ids if request.return_token_ids else None, prompt_tokens=prompt_tokens if request.return_token_ids else None, completion_tokens=output.get("completion_tokens") if request.return_token_ids else None, ) if response_processor.enable_multimodal_content(): message.multimodal_content = output.get("multipart") else: message.content = output["text"] if output.get("audio_content", None) is not None: message.audio_content = output["audio_content"] logprobs_full_res = None draft_logprobs_full_res = None prompt_logprobs_full_res = None if logprob_contents[idx]: logprobs_full_res = LogProbs(content=logprob_contents[idx]) if draft_logprob_contents[idx]: draft_logprobs_full_res = LogProbs(content=draft_logprob_contents[idx]) if prompt_logprobs_res_list[idx]: prompt_logprobs_full_res = prompt_logprobs_res_list[idx] num_cached_tokens[idx] = data.get("num_cached_tokens", 0) num_input_image_tokens[idx] = data.get("num_input_image_tokens", 0) num_input_video_tokens[idx] = data.get("num_input_video_tokens", 0) num_image_tokens[idx] = output.get("num_image_tokens", 0) finish_reason = "stop" if previous_num_tokens != max_tokens: finish_reason = "stop" if output.get("tool_call"): finish_reason = "tool_calls" else: finish_reason = "length" if data.get("error_msg") is not None and "Recover" in data["error_msg"]: finish_reason = "recover_stop" return ChatCompletionResponseChoice( index=idx, message=message, logprobs=logprobs_full_res, draft_logprobs=draft_logprobs_full_res, prompt_logprobs=prompt_logprobs_full_res, finish_reason=finish_reason, speculate_metrics=speculate_metrics, ) def _create_chat_logprobs( self, output_top_logprobs, request_logprobs: Optional[bool] = None, request_top_logprobs: Optional[int] = None, request_decode_flag: Optional[bool] = True, ) -> Optional[LogProbs]: """Create OpenAI-style logprobs for chat completions.""" if output_top_logprobs is None or len(output_top_logprobs) < 3 or any(not lst for lst in output_top_logprobs): return None logprobs_res: Optional[LogProbs] = None for logprob_token_ids, logprobs, sampled_token_ranks in zip( output_top_logprobs[0], output_top_logprobs[1], output_top_logprobs[2] ): top_logprobs = LogprobsLists( logprob_token_ids=[logprob_token_ids], logprobs=[logprobs], sampled_token_ranks=[sampled_token_ranks], ) step_logprobs_res = self._build_logprobs_response( request_logprobs=request_logprobs, response_logprobs=top_logprobs, request_top_logprobs=request_top_logprobs, request_decode_flag=request_decode_flag, ) if logprobs_res is None: logprobs_res = step_logprobs_res else: logprobs_res.content.extend(step_logprobs_res.content) return logprobs_res def _build_logprobs_response( self, request_logprobs: bool, response_logprobs: Optional[LogprobsLists], request_top_logprobs: int, request_decode_flag: bool, ) -> Optional[LogProbs]: """ Construct a logprobs response object in line with the OpenAI style. Retain the complete top-k candidates and avoid circular references. """ # Parameter validation if ( response_logprobs is None or not request_logprobs or request_top_logprobs is None or request_top_logprobs < 0 ): return None try: # The top-k candidates for the current token topk_token_ids = [] topk_logprobs = [] if response_logprobs.logprob_token_ids and len(response_logprobs.logprob_token_ids) > 0: topk_token_ids = response_logprobs.logprob_token_ids[0][: request_top_logprobs + 1] if response_logprobs.logprobs and len(response_logprobs.logprobs) > 0: topk_logprobs = response_logprobs.logprobs[0][: request_top_logprobs + 1] # Construct the candidate token structure (LogProbEntry) of topk top_logprob_entries: List[LogProbEntry] = [] for tid, lp in zip(topk_token_ids, topk_logprobs): if request_decode_flag: token_str = self.engine_client.data_processor.process_logprob_response( [tid], clean_up_tokenization_spaces=False ) token_bytes = token_str.encode("utf-8", errors="replace") if "\ufffd" in token_str: token_str = "bytes:" + "".join(f"\\x{byte:02x}" for byte in token_bytes) else: token_str = "" token_bytes = [] entry = LogProbEntry(token=token_str, logprob=lp, bytes=list(token_bytes)) top_logprob_entries.append(entry) # Construct the sampled token object (avoid sharing references with top_logprob_entries) sampled_entry = LogProbEntry( token=top_logprob_entries[0].token, logprob=top_logprob_entries[0].logprob, bytes=top_logprob_entries[0].bytes, top_logprobs=top_logprob_entries[1:], # Here are the complete topk candidates ) return LogProbs(content=[sampled_entry]) except Exception as e: error_msg = f"Error in _build_logprobs_response: {e}, {str(traceback.format_exc())}" api_server_logger.error(error_msg) return None def _get_thinking_status(self, request: ChatCompletionRequest) -> bool: """ Get the thinking status from the request. """ enable_thinking = request.chat_template_kwargs.get("enable_thinking") if request.chat_template_kwargs else None if enable_thinking is None: enable_thinking = request.metadata.get("enable_thinking") if request.metadata else None options = request.chat_template_kwargs.get("options") if request.chat_template_kwargs else None if options: thinking_mode = options.get("thinking_mode") if thinking_mode: if thinking_mode == "close" or thinking_mode == "false": enable_thinking = False else: enable_thinking = True return enable_thinking def _build_prompt_logprobs( self, prompt_logprobs_tensors: LogprobsTensors, num_prompt_logprobs: int, include_logprobs_decode_token: bool, ): """Update with prompt logprobs from worker. Args: prompt_logprobs_tensors: tuple containing the prompt logprobs tensors. """ token_ids, logprobs, ranks = prompt_logprobs_tensors # Detokenize non-incrementally. # Output is flat: [num_tok, num_lps] -> [num_tok * num_lps] if include_logprobs_decode_token: decoded_tokens = [ self.engine_client.data_processor.process_logprob_response(token_id) for token_id in token_ids.flatten().tolist() ] else: decoded_tokens = None # Recover shapes. num_prompt_tokens, num_logprobs = logprobs.shape # Pythonize the paddle tensors. prompt_token_ranks = ranks.tolist() prompt_logprobs = logprobs.tolist() token_ids = token_ids.tolist() result: Optional[PromptLogprobs] = [None] # Make Logprob for each position. for pos in range(num_prompt_tokens): # Handle flattening. offset = pos * num_logprobs offset_end = offset + num_logprobs decoded_tokens_for_pos = NONES if decoded_tokens is None else decoded_tokens[offset:offset_end] # Update with the Logprob dictionary for this pos. result.append( self._make_logprob_dict( prompt_logprobs[pos], token_ids[pos], decoded_tokens_for_pos, prompt_token_ranks[pos], num_prompt_logprobs, ) ) return result @staticmethod def _make_logprob_dict( logprobs: list[float], logprob_token_ids: list[int], decoded_tokens: Iterable[str | None], rank: int, num_logprobs: int, ) -> dict[int, Logprob]: """Make a Logprob dictionary for a position. Args: logprobs: list of log probabilities logprob_token_ids: list of top token ids decoded_tokens: list of decoded top tokens rank: rank of the sampled token num_logprobs: number of logprobs requested by the user (in addition to sampled logprob) Returns: dict[token id, Logprob] """ if num_logprobs == -1: num_logprobs = len(logprobs) # We do not need a special case for the sampled token # being in the topk, since inserting duplicated data # into a dictionary twice is the same as doing it once. topk_ranks = range(1, num_logprobs + 1) ranks = itertools.chain((rank,), topk_ranks) return { token_id: Logprob( logprob=logprob, rank=rank, decoded_token=token, ) for token_id, logprob, rank, token in zip(logprob_token_ids, logprobs, ranks, decoded_tokens) }