Update to new version of base image

vllm/model_executor/guided_decoding/outlines_logits_processors.py

@@ -23,12 +23,21 @@ from typing import Callable, DefaultDict, Dict, List, Union
 import torch
 from lark import Lark
 from outlines import grammars
-from outlines.caching import cache
+from outlines.caching import cache, disable_cache
 from outlines.fsm.guide import CFGGuide, Generate, Guide, RegexGuide, Write
 from outlines.fsm.json_schema import build_regex_from_schema
 from pydantic import BaseModel
 from transformers import PreTrainedTokenizerBase
+import vllm.envs as envs
+from vllm.logger import init_logger
+logger = init_logger(__name__)
 
+if envs.VLLM_V0_USE_OUTLINES_CACHE:
+    logger.warning("Enabling outlines cache. This is an unbounded on-disk "
+                   "cache. It may consume a lot of disk space and should "
+                   "not be used with untrusted clients.")
+else:
+    disable_cache()
 
 class BaseLogitsProcessor: