selfhostyourtech/reflector
2
0
Fork 0
mirror of https://github.com/Monadical-SAS/reflector.git synced 2025-12-21 04:39:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

server: fixes initial timeout for llm warmup

Browse Source
This commit is contained in:
Mathieu Virbel
2023-08-11 15:45:28 +02:00
parent 38a5ee0da2
commit a06056f9bc
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
server/reflector/llm/base.py
View File

@@ -35,7 +35,7 @@ class LLM:
name = self.__class__.__name__ name = self.__class__.__name__
logger.info(f"LLM[{name}] warming up...") logger.info(f"LLM[{name}] warming up...")
try: try:
await retry(self._warmup)(logger=logger) await self._warmup(logger=logger)
duration = monotonic() - start duration = monotonic() - start
logger.info(f"LLM[{name}] warmup took {duration:.2f} seconds") logger.info(f"LLM[{name}] warmup took {duration:.2f} seconds")
except Exception: except Exception:

2
server/reflector/llm/llm_modal.py
View File

@@ -19,7 +19,7 @@ class ModalLLM(LLM):
response = await client.post( response = await client.post(
self.llm_warmup_url, self.llm_warmup_url,
headers=self.headers, headers=self.headers,
timeout=self.timeout, timeout=60**5,
) )
response.raise_for_status() response.raise_for_status()