[Chore] Remove Sampler from Model Code (#17084)

Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>

vllm/model_executor/models/gemma3_mm.py

@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: Apache-2.0
 import math
 from collections.abc import Iterable, Mapping, Sequence
-from typing import Any, Literal, Optional, Set, Tuple, TypedDict, Union
+from typing import Any, Literal, Optional, Set, Tuple, TypedDict
 
 import torch
 from torch import nn
@@ -12,7 +12,6 @@ import vllm.envs as envs
 from vllm.config import VllmConfig
 from vllm.logger import init_logger
 from vllm.model_executor.layers.layernorm import GemmaRMSNorm
-from vllm.model_executor.layers.sampler import SamplerOutput
 from vllm.model_executor.models.module_mapping import MultiModelKeys
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.multimodal import MULTIMODAL_REGISTRY
@@ -503,10 +502,6 @@ class Gemma3ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
     def dtype(self):
         return next(self.parameters()).dtype
 
-    @property
-    def sampler(self):
-        return self.language_model.sampler
-
     def _validate_pixel_values(self, data: torch.Tensor) -> torch.Tensor:
         h = w = self.config.vision_config.image_size
         expected_dims = (3, h, w)
@@ -607,7 +602,7 @@ class Gemma3ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
                 positions: torch.Tensor,
                 intermediate_tensors: Optional[IntermediateTensors] = None,
                 inputs_embeds: Optional[torch.Tensor] = None,
-                **kwargs: object) -> Union[SamplerOutput, IntermediateTensors]:
+                **kwargs: object) -> IntermediateTensors:
         if intermediate_tensors is not None:
             inputs_embeds = None
 
@@ -704,13 +699,6 @@ class Gemma3ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
         return self.language_model.compute_logits(hidden_states,
                                                   sampling_metadata)
 
-    def sample(
-        self,
-        logits: torch.Tensor,
-        sampling_metadata: SamplingMetadata,
-    ) -> Optional[SamplerOutput]:
-        return self.language_model.sample(logits, sampling_metadata)
-
     def load_weights(self, weights: Iterable[Tuple[str,
                                                    torch.Tensor]]) -> Set[str]:
         loader = AutoWeightsLoader(self)