fix run error

vllm/model_executor/layers/sampler.py

@@ -125,6 +125,16 @@ class SamplerOutput(
     # Time taken in the model execute function. This will include model forward,
     # block/sync across workers, cpu-gpu sync time and sampling time.
     model_execute_time: Optional[float] = None
+    
+    # Optional lm_head logits from the model.
+    logits: Optional[torch.Tensor] = None
+
+    # tree-style cartesian candidates
+    cart_candidates: Optional[torch.Tensor] = None
+
+    # tree-style cartesian candidates
+    tree_attn_masks: Optional[torch.Tensor] = None
+
 
     def __getitem__(self, idx: int) -> CompletionSequenceGroupOutput:
         return self.outputs[idx]

vllm/model_executor/models/chatglm.py

@@ -511,8 +511,6 @@ class ChatGLMModel(nn.Module):
 
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
-        
-        self.config = config
 
         config = vllm_config.model_config.hf_config
         cache_config = vllm_config.cache_config

vllm/platforms/__init__.py

 from .interface import _Backend  # noqa: F401
 from .interface import CpuArchEnum, Platform, PlatformEnum, UnspecifiedPlatform
+import torch
 
 current_platform: Platform
 

vllm/worker/model_runner.py

@@ -476,7 +476,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             self.block_aligned_sliding_window = \
                 self.sliding_window_blocks * self.block_size
 
-        self.is_encoder_decoder_model = self.runner.model_config.is_encoder_decoder_model
+        self.is_encoder_decoder_model = self.runner.model_config.is_encoder_decoder
 
     def _compute_lens(self, inter_data: InterDataForSeqGroup, seq_idx: int,
                       seq_group_metadata: SequenceGroupMetadata):