[CI/Build] mypy: Resolve some errors from checking vllm/engine (#9267)
Signed-off-by: Russell Bryant <rbryant@redhat.com>
This commit is contained in:
@@ -3,7 +3,7 @@ import copy
|
||||
import pickle
|
||||
from contextlib import contextmanager, suppress
|
||||
from typing import (Any, AsyncGenerator, Dict, Iterator, List, Mapping,
|
||||
Optional, Union, overload)
|
||||
Optional, Union, cast, overload)
|
||||
|
||||
import cloudpickle
|
||||
import zmq
|
||||
@@ -513,9 +513,14 @@ class MQLLMEngineClient(EngineClient):
|
||||
assert (prompt is not None and pooling_params is not None
|
||||
and request_id is not None)
|
||||
|
||||
return self._process_request(prompt, pooling_params, request_id,
|
||||
lora_request, trace_headers, None,
|
||||
priority)
|
||||
return cast(
|
||||
AsyncGenerator[EmbeddingRequestOutput, None],
|
||||
self._process_request(prompt,
|
||||
pooling_params,
|
||||
request_id,
|
||||
lora_request,
|
||||
trace_headers,
|
||||
priority=priority))
|
||||
|
||||
async def _process_request(
|
||||
self,
|
||||
@@ -543,7 +548,9 @@ class MQLLMEngineClient(EngineClient):
|
||||
build_guided_decoding_logits_processor_async(
|
||||
sampling_params=params,
|
||||
tokenizer=await self.get_tokenizer(lora_request),
|
||||
default_guided_backend=self.decoding_config.guided_decoding_backend
|
||||
default_guided_backend=(self.decoding_config.guided_decoding_backend
|
||||
if self.decoding_config
|
||||
else DecodingConfig.guided_decoding_backend),
|
||||
)
|
||||
|
||||
# 1) Create output queue for this requests.
|
||||
|
||||
Reference in New Issue
Block a user