[CI] Update vllm version to 20250922(5aeb925) (#3091)

### What this PR does / why we need it?
This pr bump vllm commit hash to
5aeb925452
fix issues:  
1. https://github.com/vllm-project/vllm/pull/25345 has remove v0
metadata
2. https://github.com/vllm-project/vllm/pull/25332
3. https://github.com/vllm-project/vllm/pull/25334
4. https://github.com/vllm-project/vllm/pull/23558, note that this vllm
commit update the model register logic, which will check all the model
registered have the `vllm.model_executor.models` path , which breaks our
custom registration of the deepseek_v3 model (it doesn't exist in the
vllm model path). so I move deepseek_v3 model registy to deepseek_v2 to
solve temporary

### How was this patch tested?

- vLLM version: v0.10.2
- vLLM main:
9607d5eb44

---------

Signed-off-by: wangli <wangli858794774@gmail.com>

vllm_ascend/torchair/models/qwen2.py

@@ -40,7 +40,6 @@ from vllm.model_executor.models.qwen2 import Qwen2ForCausalLM  # noqa: F401
 from vllm.model_executor.models.qwen2 import Qwen2MLP, Qwen2Model
 from vllm.model_executor.models.utils import (AutoWeightsLoader,
                                               PPMissingLayer, maybe_prefix)
-from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors
 
 from vllm_ascend.ascend_config import get_ascend_config
@@ -343,9 +342,9 @@ class CustomQwen2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
         return hidden_states
 
     def compute_logits(
-        self,
-        hidden_states: torch.Tensor,
-        sampling_metadata: SamplingMetadata,
+            self,
+            hidden_states: torch.Tensor,
+            sampling_metadata,  # type: ignore
     ) -> Optional[torch.Tensor]:
         logits = self.logits_processor(self.lm_head, hidden_states,
                                        sampling_metadata)

vllm_ascend/torchair/models/torchair_deepseek_mtp.py

@@ -27,14 +27,12 @@ from vllm.config import CacheConfig, ModelConfig, VllmConfig
 from vllm.model_executor.layers.layernorm import RMSNorm
 from vllm.model_executor.layers.logits_processor import LogitsProcessor
 from vllm.model_executor.layers.quantization import QuantizationConfig
-from vllm.model_executor.layers.sampler import get_sampler
 from vllm.model_executor.layers.vocab_parallel_embedding import (
     ParallelLMHead, VocabParallelEmbedding)
 from vllm.model_executor.models.deepseek_mtp import (
     DeepSeekMTP, DeepSeekMultiTokenPredictor, DeepSeekMultiTokenPredictorLayer,
     SharedHead)
 from vllm.model_executor.models.utils import maybe_prefix
-from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors
 
 from vllm_ascend.torchair.models.torchair_deepseek_v2 import \
@@ -172,7 +170,7 @@ class TorchairDeepSeekMultiTokenPredictor(DeepSeekMultiTokenPredictor):
     def compute_logits(
         self,
         hidden_states: torch.Tensor,
-        sampling_metadata: SamplingMetadata,
+        sampling_metadata,  # type: ignore
         spec_step_idx: int = 0,
     ) -> torch.Tensor:
         current_step_idx = (spec_step_idx % self.num_mtp_layers)
@@ -199,8 +197,6 @@ class TorchairDeepSeekMTP(DeepSeekMTP):
         self.model = TorchairDeepSeekMultiTokenPredictor(
             vllm_config=vllm_config, prefix=maybe_prefix(prefix, "model"))
 
-        self.sampler = get_sampler()
-
     def forward(
         self,
         input_ids: torch.Tensor,

vllm_ascend/torchair/models/torchair_deepseek_v2.py

@@ -52,7 +52,6 @@ from vllm.model_executor.layers.linear import (ColumnParallelLinear,
 from vllm.model_executor.layers.logits_processor import LogitsProcessor
 from vllm.model_executor.layers.quantization import QuantizationConfig
 from vllm.model_executor.layers.rotary_embedding import get_rope
-from vllm.model_executor.layers.sampler import get_sampler
 from vllm.model_executor.layers.vocab_parallel_embedding import (
     ParallelLMHead, VocabParallelEmbedding)
 from vllm.model_executor.model_loader.weight_utils import (
@@ -945,7 +944,6 @@ class TorchairDeepseekV2ForCausalLM(DeepseekV2ForCausalLM):
         else:
             self.lm_head = PPMissingLayer()
         self.logits_processor = LogitsProcessor(config.vocab_size)
-        self.sampler = get_sampler()
         self.make_empty_intermediate_tensors = (
             self.model.make_empty_intermediate_tensors)
 

vllm_ascend/torchair/models/torchair_pangu_moe.py

@@ -45,7 +45,6 @@ from vllm.model_executor.layers.linear import (LinearBase,
 from vllm.model_executor.layers.logits_processor import LogitsProcessor
 from vllm.model_executor.layers.quantization import QuantizationConfig
 from vllm.model_executor.layers.rotary_embedding import get_rope
-from vllm.model_executor.layers.sampler import SamplerOutput, get_sampler
 from vllm.model_executor.layers.vocab_parallel_embedding import (
     ParallelLMHead, VocabParallelEmbedding)
 from vllm.model_executor.model_loader.weight_utils import default_weight_loader
@@ -53,9 +52,9 @@ from vllm.model_executor.models.interfaces import SupportsPP
 from vllm.model_executor.models.utils import (
     extract_layer_index, is_pp_missing_parameter,
     make_empty_intermediate_tensors_factory, make_layers, maybe_prefix)
-from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.model_executor.utils import set_weight_attrs
 from vllm.sequence import IntermediateTensors
+from vllm.v1.sample.sampler import Sampler
 
 from vllm_ascend.ascend_config import get_ascend_config
 from vllm_ascend.utils import ACL_FORMAT_FRACTAL_NZ, is_310p
@@ -913,7 +912,7 @@ class PanguProMoEForCausalLM(nn.Module, SupportsPP):
         if self.config.tie_word_embeddings:
             self.lm_head.weight = self.model.embed_tokens.weight
         self.logits_processor = LogitsProcessor(config.vocab_size)
-        self.sampler = get_sampler()
+        self.sampler = Sampler()
         self.make_empty_intermediate_tensors = (
             self.model.make_empty_intermediate_tensors)
 
@@ -935,19 +934,19 @@ class PanguProMoEForCausalLM(nn.Module, SupportsPP):
         return hidden_states
 
     def compute_logits(
-        self,
-        hidden_states: torch.Tensor,
-        sampling_metadata: SamplingMetadata,
+            self,
+            hidden_states: torch.Tensor,
+            sampling_metadata,  # type: ignore
     ) -> Optional[torch.Tensor]:
         logits = self.logits_processor(self.lm_head, hidden_states,
                                        sampling_metadata)
         return logits
 
     def sample(
-        self,
-        logits: Optional[torch.Tensor],
-        sampling_metadata: SamplingMetadata,
-    ) -> Optional[SamplerOutput]:
+            self,
+            logits: Optional[torch.Tensor],
+            sampling_metadata,  # type: ignore
+    ):
         next_tokens = self.sampler(logits, sampling_metadata)
         return next_tokens