biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Frontend] Use new Renderer for Completions and Tokenize API (#32863)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-31 20:51:15 +08:00
committed by GitHub
parent 8980001c93
commit f0a1c8453a
64 changed files with 2116 additions and 2003 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/utils/async_utils.py
View File

@@ -50,14 +50,17 @@ class AsyncMicrobatchTokenizer:
self._executor = ThreadPoolExecutor(max_workers=1)
# === Public async API ===
async def __call__(self, prompt, **kwargs):
async def __call__(self, prompt, **kwargs) -> BatchEncoding:
result_future: Future = self._loop.create_future()
key = self._queue_key("encode", kwargs)
queue = self._get_queue(self._loop, key)
await queue.put((prompt, kwargs, result_future))
return await result_future
async def decode(self, token_ids, **kwargs):
async def encode(self, prompt, **kwargs) -> list[int]:
return (await self(prompt, **kwargs)).input_ids
async def decode(self, token_ids, **kwargs) -> str:
result_future: Future = self._loop.create_future()
key = self._queue_key("decode", kwargs)
queue = self._get_queue(self._loop, key)