[Frontend] Add --spec-method/--spec-model/--spec-tokens CLI aliases (#42476)

Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
This commit is contained in:
Michael Goin
2026-05-18 17:22:27 -07:00
committed by GitHub
co-authored by Claude
parent f85c76d701
commit 239b5ff30c
2 changed files with 34 additions and 0 deletions
+10
View File
@@ -188,6 +188,10 @@ class LLM(PoolingOfflineMixin):
dictionary or an AttentionConfig instance. If a dictionary, it will
be converted to an AttentionConfig. Allows specifying the attention
backend and other attention-related settings.
spec_method: Top-level alias for `speculative_config["method"]`.
spec_model: Top-level alias for `speculative_config["model"]`.
spec_tokens: Top-level alias for
`speculative_config["num_speculative_tokens"]`.
**kwargs: Arguments for [`EngineArgs`][vllm.EngineArgs].
Note:
@@ -236,6 +240,9 @@ class LLM(PoolingOfflineMixin):
compilation_config: int | dict[str, Any] | CompilationConfig | None = None,
quantization_config: dict[str, Any] | QuantizationConfigArgs | None = None,
logits_processors: list[str | type[LogitsProcessor]] | None = None,
spec_method: str | None = None,
spec_model: str | None = None,
spec_tokens: int | None = None,
**kwargs: Any,
) -> None:
"""LLM constructor."""
@@ -357,6 +364,9 @@ class LLM(PoolingOfflineMixin):
compilation_config=compilation_config_instance,
quantization_config=quantization_config,
logits_processors=logits_processors,
spec_method=spec_method,
spec_model=spec_model,
spec_tokens=spec_tokens,
**kwargs,
)