增加medusa并行解码功能,后续增加使用说明和测试文档

vllm/worker/model_runner.py

@@ -13,6 +13,7 @@ import numpy as np
 import torch
 import torch.distributed
 import torch.nn as nn
+import torch.nn.functional as F
 
 import vllm.envs as envs
 from vllm.attention import AttentionMetadata, get_attn_backend
@@ -197,6 +198,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             self.lora_requests.clear()  # type: ignore
             self.prompt_adapter_index_mapping.clear()  # type: ignore
             self.prompt_adapter_prompt_mapping.clear()  # type: ignore
+            self.tree_attn_masks[0] = None  # type: ignore
 
         def __init__(
             self,
@@ -244,6 +246,9 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             reinit: bool = False,
             reinit_use_defaults: bool = False,
             encoder_seq_len: int = 0,
+
+            # attention mask used in tree-style generation
+            tree_attn_masks: Optional[List[torch.Tensor]] = None,
         ):
             if reinit:
                 assert len(self.seq_ids) == len(seq_ids)  # type: ignore
@@ -335,6 +340,11 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
                     else:
                         self.prompt_adapter_prompt_mapping.clear()
 
+                    if tree_attn_masks:
+                        self.tree_attn_masks = tree_attn_masks
+                    else:
+                        self.tree_attn_masks.clear()
+
             else:
                 self.input_tokens = input_tokens or []
                 self.input_positions = input_positions or []
@@ -354,6 +364,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
                     prompt_adapter_index_mapping or [])
                 self.prompt_adapter_prompt_mapping = (
                     prompt_adapter_prompt_mapping or [])
+                self.tree_attn_masks = tree_attn_masks or []
 
             self.prompt_adapter_request = prompt_adapter_request
             self.multi_modal_inputs = multi_modal_inputs
@@ -369,6 +380,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
 
             self.input_tokens = [[] for _ in range(self.n_seqs)]
             self.input_positions = [[] for _ in range(self.n_seqs)]
+            self.tree_attn_masks = [None for _ in range(self.n_seqs)]
             self.mrope_input_positions = None
             self.seq_lens = [0] * self.n_seqs
             self.orig_seq_lens = [0] * self.n_seqs
@@ -502,6 +514,16 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
         else:
             inter_data.input_positions[seq_idx].extend(
                 range(context_len, seq_len))
+            
+        if seq_group_metadata.tree_position_ids is None:
+            if (seq_len - context_len) == 1:
+                inter_data.input_positions[seq_idx].append(seq_len - 1)
+            else:
+                inter_data.input_positions[seq_idx].extend(
+                    range(context_len, seq_len))
+        else:
+            inter_data.input_positions[seq_idx] = seq_group_metadata.tree_position_ids.cpu().tolist()
+            inter_data.tree_attn_masks[seq_idx] = seq_group_metadata.tree_attn_masks
 
         inter_data.query_lens[
             seq_idx] = seq_len - context_len if inter_data.is_prompt else 1
@@ -835,10 +857,25 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
         # Sequence and query lengths.
         if cuda_graph_pad_size:
             seq_lens.extend(itertools.repeat(1, cuda_graph_pad_size))
+        
+        # prepare tree attention masks
+        max_context_len = 0
+        for inter_data in self.inter_data_list:
+            max_context_len = max(max_context_len, max(inter_data.context_lens))
+        tree_attention_masks_list = []
+        for inter_data in self.inter_data_list:
+            for i in range(len(inter_data.seq_lens)):
+                tree_attn_masks = inter_data.tree_attn_masks[i]
+                if tree_attn_masks is not None:
+                    tree_attention_masks_list.append(tree_attn_masks)
+        tree_attention_masks_tensor = None
+        if len(tree_attention_masks_list) > 0:
+            tree_attention_masks_tensor = torch.stack(tree_attention_masks_list, dim=0)
 
         # Attention metadata.
         attn_metadata = self.attn_metadata_builder.build(
-            seq_lens, query_lens, cuda_graph_pad_size, batch_size)
+            seq_lens, query_lens, cuda_graph_pad_size, batch_size,
+            tree_attention_masks_tensor=tree_attention_masks_tensor)
 
         # LoRA data.
         lora_requests = set()

vllm/worker/neuron_worker.py

@@ -10,6 +10,7 @@ from vllm.distributed import (ensure_model_parallel_initialized,
                               init_distributed_environment)
 from vllm.model_executor import set_random_seed
 from vllm.sequence import ExecuteModelRequest
+from vllm.worker.cache_engine import CacheEngine
 from vllm.worker.neuron_model_runner import NeuronModelRunner
 from vllm.worker.worker_base import (LocalOrDistributedWorkerBase,
                                      LoraNotSupportedWorkerBase, WorkerInput)
@@ -92,6 +93,10 @@ class NeuronWorker(LoraNotSupportedWorkerBase, LocalOrDistributedWorkerBase):
     @property
     def kv_cache(self) -> Optional[List[List[torch.Tensor]]]:
         return None
+    
+    @property
+    def cache_engines(self) -> Optional[List[CacheEngine]]:
+        return None
 
     @torch.inference_mode()
     def prepare_worker_input(

vllm/worker/tpu_worker.py

@@ -15,6 +15,7 @@ from vllm.model_executor import set_random_seed
 from vllm.sequence import ExecuteModelRequest
 from vllm.utils import STR_DTYPE_TO_TORCH_DTYPE, get_dtype_size
 from vllm.worker.tpu_model_runner import TPUModelRunner
+from vllm.worker.cache_engine import CacheEngine
 from vllm.worker.worker_base import (LocalOrDistributedWorkerBase,
                                      LoraNotSupportedWorkerBase, WorkerInput)
 
@@ -212,6 +213,10 @@ class TPUWorker(LoraNotSupportedWorkerBase, LocalOrDistributedWorkerBase):
         # NOTE(woosuk): This assumes virtual_engine == 0, i.e., no pipeline
         # parallelism.
         return [self.tpu_cache]
+    
+    @property
+    def cache_engines(self) -> Optional[List[CacheEngine]]:
+        return None
 
     def prepare_worker_input(
         self,

vllm/worker/worker.py

@@ -291,6 +291,10 @@ class Worker(LocalOrDistributedWorkerBase):
     @property
     def kv_cache(self) -> Optional[List[List[torch.Tensor]]]:
         return self.gpu_cache
+    
+    @property
+    def cache_engines(self) -> Optional[List[CacheEngine]]:
+        return self.cache_engine
 
     @torch.inference_mode()
     def prepare_worker_input(

vllm/worker/worker_base.py

@@ -16,6 +16,7 @@ from vllm.platforms import current_platform
 from vllm.sequence import ExecuteModelRequest, IntermediateTensors
 from vllm.utils import (enable_trace_function_call_for_thread,
                         update_environment_variables)
+from vllm.worker.cache_engine import CacheEngine
 from vllm.worker.model_runner_base import (BroadcastableModelInput,
                                            ModelRunnerBase,
                                            ModelRunnerInputBase)
@@ -29,6 +30,8 @@ class WorkerBase(ABC):
     communicate request metadata to other workers.
     """
 
+    model_input: Optional[ModelRunnerInputBase] = None
+
     @abstractmethod
     def init_device(self) -> None:
         """Initialize device state, such as loading the model or other on-device
@@ -99,6 +102,23 @@ class WorkerBase(ABC):
     @abstractmethod
     def list_loras(self) -> Set[int]:
         raise NotImplementedError
+    
+    @property
+    @abstractmethod
+    def kv_cache(self) -> Optional[List[List[torch.Tensor]]]:
+        """
+        Gets the list of kv caches to pass to the worker's model runner. Each
+        element in the list is a kv cache corresponding to a particular virtual
+        engine (PP stream). Used by the default `execute_model`. If the worker's
+        model runner does not follow the ModelRunnerBase interface, then inherit
+        from WorkerBase instead.
+        """
+        raise NotImplementedError
+    
+    @property
+    @abstractmethod
+    def cache_engines(self) -> Optional[List[CacheEngine]]:
+        raise NotImplementedError
 
 
 class LoraNotSupportedWorkerBase(WorkerBase):
@@ -118,6 +138,14 @@ class LoraNotSupportedWorkerBase(WorkerBase):
 
     def list_loras(self) -> Set[int]:
         raise ValueError(f"{type(self)} does not support LoRA")
+    
+    @property
+    def kv_cache(self) -> Optional[List[List[torch.Tensor]]]:
+        return None
+
+    @property
+    def cache_engines(self) -> Optional[List[CacheEngine]]:
+        return None
 
 
 @dataclasses.dataclass(frozen=True)
@@ -249,6 +277,15 @@ class LocalOrDistributedWorkerBase(WorkerBase):
 
         worker_input: WorkerInput = self.prepare_worker_input(
             execute_model_req=execute_model_req)
+        
+        # set tree_attn_masks and position ids to seq_group_metadata_list
+        if execute_model_req.tree_attn_masks is not None:
+            for i, seq_group_metadata in enumerate(execute_model_req.seq_group_metadata_list):
+                # seq_group_metadata.tree_attn_masks = execute_model_req.tree_attn_masks[i]
+                # seq_group_metadata.tree_position_ids = execute_model_req.tree_position_ids[i]
+                seq_group_metadata.set_tree_style_args(tree_attn_masks=execute_model_req.tree_attn_masks[i], 
+                                                       tree_position_ids=execute_model_req.tree_position_ids[i])
+
         model_input: ModelRunnerInputBase = (
             self.model_runner.prepare_model_input(
                 execute_model_req.seq_group_metadata_list,
@@ -307,6 +344,8 @@ class LocalOrDistributedWorkerBase(WorkerBase):
         model_input, worker_input, kwargs = inputs
         num_steps = worker_input.num_steps
 
+        self.model_input = model_input
+
         self.execute_worker(worker_input)
 
         # If there is no input, we don't need to execute the model.