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

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

vllm/model_executor/models/blip2.py

@@ -1,7 +1,6 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from collections.abc import Iterable, Mapping, Sequence
-from functools import cached_property
 from typing import Literal, Optional, Set, Tuple, TypedDict, Union
 
 import torch
@@ -12,7 +11,6 @@ from transformers import (BatchFeature, Blip2Config, Blip2QFormerConfig,
 from vllm.config import CacheConfig, VllmConfig
 from vllm.model_executor.layers.activation import get_act_fn
 from vllm.model_executor.layers.quantization import QuantizationConfig
-from vllm.model_executor.layers.sampler import SamplerOutput, get_sampler
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.multimodal import MULTIMODAL_REGISTRY
 from vllm.multimodal.inputs import (MultiModalDataDict, MultiModalFieldConfig,
@@ -530,13 +528,6 @@ class Blip2ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
         self.make_empty_intermediate_tensors = (
             self.language_model.make_empty_intermediate_tensors)
 
-    @cached_property
-    def sampler(self):
-        if hasattr(self.language_model, "sampler"):
-            return self.language_model.sampler
-
-        return get_sampler()
-
     def _validate_pixel_values(self, data: torch.Tensor) -> torch.Tensor:
         h = w = self.config.vision_config.image_size
         expected_dims = (3, h, w)
@@ -649,7 +640,7 @@ class Blip2ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
         intermediate_tensors: Optional[IntermediateTensors] = None,
         inputs_embeds: Optional[torch.Tensor] = None,
         **kwargs: object,
-    ) -> Union[SamplerOutput, IntermediateTensors]:
+    ) -> IntermediateTensors:
         """Run forward pass for BLIP-2.
 
         One key thing to understand is the `input_ids` already accounts for the
@@ -707,13 +698,6 @@ class Blip2ForConditionalGeneration(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)