[Core] Pipeline Parallel Support (#4412)

Signed-off-by: Muralidhar Andoorveedu <muralidhar.andoorveedu@centml.ai>

vllm/model_executor/models/arctic.py

@@ -29,7 +29,7 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
 from vllm.model_executor.model_loader.weight_utils import default_weight_loader
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.model_executor.utils import set_weight_attrs
-from vllm.sequence import SamplerOutput
+from vllm.sequence import IntermediateTensors, SamplerOutput
 from vllm.transformers_utils.configs.arctic import ArcticConfig
 
 logger = init_logger(__name__)
@@ -426,6 +426,7 @@ class ArcticForCausalLM(nn.Module):
         positions: torch.Tensor,
         kv_caches: List[torch.Tensor],
         attn_metadata: AttentionMetadata,
+        intermediate_tensors: Optional[IntermediateTensors] = None,
     ) -> torch.Tensor:
         hidden_states = self.model(input_ids, positions, kv_caches,
                                    attn_metadata)