[Mypy] Fix mypy typing for Ultravox and Unlimited-OCR models (#58239)

Signed-off-by: Ashraf Bhuiyan <mbhuiyan@redhat.com>
This commit is contained in:
Ashraf Bhuiyan
2026-09-27 21:19:04 -04:00
committed by GitHub
parent 39d49539de
commit b9e2903b89
6 changed files with 36 additions and 27 deletions
@@ -1,10 +1,9 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
from collections.abc import Iterable
from collections.abc import Callable
from functools import wraps
from types import FunctionType
from typing import TYPE_CHECKING
from typing import TYPE_CHECKING, Concatenate, ParamSpec, TypeVar
import torch
@@ -20,13 +19,18 @@ if TYPE_CHECKING:
__all__ = ["set_torchao_reload_attrs", "support_quantized_model_reload_from_hp_weights"]
_P = ParamSpec("_P")
_R = TypeVar("_R")
def set_torchao_reload_attrs(model: torch.nn.Module, model_config: ModelConfig):
model._do_torchao_reload = True
model._model_config = model_config
def support_quantized_model_reload_from_hp_weights(original_load_weights: FunctionType):
def support_quantized_model_reload_from_hp_weights(
original_load_weights: Callable[Concatenate["AutoWeightsLoader", _P], _R],
) -> Callable[Concatenate["AutoWeightsLoader", _P], _R]:
"""Decorator for `load_weights` method for AutoWeightsLoader.load_weights to support
reloading high precision (bfloat16/float16/float32) weight for an already quantized
model, this involves restoring the weights to a high precision weights and
@@ -39,17 +43,16 @@ def support_quantized_model_reload_from_hp_weights(original_load_weights: Functi
@wraps(original_load_weights)
def patched_model_load_weights(
self: "AutoWeightsLoader",
weights: Iterable[tuple[str, torch.Tensor]],
*args,
**kwargs,
):
*args: _P.args,
**kwargs: _P.kwargs,
) -> _R:
model = self.module
if not getattr(model, "_do_torchao_reload", False):
return original_load_weights(self, weights, *args, **kwargs)
return original_load_weights(self, *args, **kwargs)
initialize_layerwise_reload(model)
loaded_weights = original_load_weights(self, weights, *args, **kwargs)
loaded_weights = original_load_weights(self, *args, **kwargs)
finalize_layerwise_reload(model, model._model_config)
return loaded_weights