chore: bump deps and lib version

.pre-commit-config.yaml

@@ -1,6 +1,6 @@
 repos:
 - repo: https://github.com/astral-sh/ruff-pre-commit
-  rev: v0.6.8
+  rev: v0.7.0
   hooks:
     - id: ruff
       args: [ --fix ]

decoding/models.py

@@ -24,6 +24,7 @@
 _logger.info("Importing vLLM: This may take a moment...")
 
 from vllm import LLM, SamplingParams
+from vllm.inputs import PromptType
 from vllm.outputs import RequestOutput
 from vllm.sequence import Logprob
 from vllm.transformers_utils.tokenizer import AnyTokenizer
@@ -55,6 +56,7 @@ class ModelParams(TypedDict, total=False):
     max_seq_len_to_capture: int
     disable_custom_all_reduce: bool
     disable_async_output_proc: bool
+    enable_prefix_caching: bool
 
 
 @dataclass(frozen=True, kw_only=True)
@@ -269,11 +271,11 @@ def _is_valid_logp(
     raise ValueError(msg)
 
 
-def _guard_prompt_text(x: str | None) -> str:
-    if x is None:
-        msg = "Prompt text should not be None"
-        raise ValueError(msg)
-    return x
+def _guard_prompt_text(x: PromptType | None) -> str:
+    if isinstance(x, str):  # `PromptType` is a union type of `str` and other types
+        return x
+    msg = "Prompt text should be a string"
+    raise ValueError(msg)
 
 
 def _guard_output_logp(x: float | None) -> float:

setup.py

@@ -9,24 +9,24 @@
 core_requirements = [
     "jax==0.4.31",
     "jaxtyping==0.2.34",
-    "vllm==0.6.2",
+    "vllm==0.6.3.post1",
 ]
 dev_requirements = [
-    "pdoc==14.7.0",
-    "pre-commit==3.8.0",
-    "pyright==1.1.382.post1",
+    "pdoc==15.0.0",
+    "pre-commit==4.0.1",
+    "pyright==1.1.385",
     "pytest==8.3.3",
     "pytest-cov==5.0.0",
     "pytest-html==4.1.1",
-    "ruff==0.6.8",
+    "ruff==0.7.0",
 ]
 
 with pathlib.Path("README.md").open(encoding="utf-8") as f:
     readme = f.read()
 
 setuptools.setup(
     name="decoding",
-    version="0.1.1",
+    version="0.1.2",
     description="Composable inference algorithms with LLMs and programmable logic",
     long_description=readme,
     long_description_content_type="text/markdown",