[Frontend] Add the Beam reasoning and tool parsers (#62052)

Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
Co-authored-by: Bugen Zhao <i@bugenzhao.com>
Co-authored-by: Yoni Gozlan <74535834+yonigozlan@users.noreply.github.com>
Co-authored-by: neilkg <33635204+neilkg@users.noreply.github.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Woosuk Kwon
2026-10-16 14:08:29 -07:00
committed by GitHub
co-authored by Bugen Zhao Yoni Gozlan neilkg Cursor Claude Opus 5.5
parent 9731f30a03
commit 153a17166e
27 changed files with 4464 additions and 92 deletions
+6
View File
@@ -246,6 +246,7 @@ if TYPE_CHECKING:
VLLM_MAX_TOKENS_PER_EXPERT_FP4_MOE: int = 163840
VLLM_TOOL_PARSE_REGEX_TIMEOUT_SECONDS: int = 1
VLLM_ENFORCE_STRICT_TOOL_CALLING: bool = True
VLLM_BEAM_ALWAYS_CONSTRAIN_OUTPUT: bool = True
VLLM_MQ_MAX_CHUNK_BYTES_MB: int = 16
VLLM_EXECUTE_MODEL_TIMEOUT_SECONDS: int = 300
VLLM_WORKER_SHUTDOWN_TIMEOUT_SECONDS: int = 5
@@ -1844,6 +1845,11 @@ environment_variables: dict[str, Callable[[], Any]] = {
"VLLM_ENFORCE_STRICT_TOOL_CALLING": lambda: (
os.getenv("VLLM_ENFORCE_STRICT_TOOL_CALLING", "True").lower() in ("true", "1")
),
# Constrain every `beam` output to its response template's format, also
# without tool or structured output constraints.
"VLLM_BEAM_ALWAYS_CONSTRAIN_OUTPUT": lambda: (
os.getenv("VLLM_BEAM_ALWAYS_CONSTRAIN_OUTPUT", "True").lower() in ("true", "1")
),
# Control the max chunk bytes (in MB) for the rpc message queue.
# Object larger than this threshold will be broadcast to worker
# processes via zmq.