Bump vLLM version

docs/reference/vllm.md

@@ -4,13 +4,8 @@ Outlines can be deployed as an LLM service using the vLLM inference engine and a
 
 ```bash
 pip install outlines[serve]
-pip install "pydantic>=2.0"
 ```
 
-!!! Warning
-
-    Updating Pydantic to v2 after the installation is necessary.
-
 You can then start the server with:
 
 ```bash

outlines/serve/vllm.py

@@ -51,7 +51,7 @@ def __init__(self, regex_string, llm):
             An instance of `vllm.LLM`
 
         """
-        tokenizer = self.adapt_tokenizer(llm.tokenizer)
+        tokenizer = self.adapt_tokenizer(llm.tokenizer.tokenizer)
 
         fsm = RegexFSM(regex_string, tokenizer)
         self.fsm = fsm

pyproject.toml

@@ -60,7 +60,7 @@ test = [
     "huggingface_hub"
 ]
 serve = [
-    "vllm>=0.2.6",
+    "vllm>=0.3.0",
     "ray==2.9.0",
     "uvicorn",
     "fastapi"