Change the name to vLLM (#150)

cacheflow/outputs.py → vllm/outputs.py

 from typing import Dict, List, Optional
 
-from cacheflow.sequence import SequenceGroup, SequenceStatus
+from vllm.sequence import SequenceGroup, SequenceStatus
 
 
 class CompletionOutput:

cacheflow/sequence.py → vllm/sequence.py

@@ -2,8 +2,8 @@ import copy
 import enum
 from typing import Dict, List, Optional, Union
 
-from cacheflow.block import LogicalTokenBlock
-from cacheflow.sampling_params import SamplingParams
+from vllm.block import LogicalTokenBlock
+from vllm.sampling_params import SamplingParams
 
 
 class SequenceStatus(enum.Enum):

cacheflow/worker/cache_engine.py → vllm/worker/cache_engine.py

@@ -3,8 +3,8 @@ from typing import Dict, List, Tuple
 
 import torch
 
-from cacheflow import cache_ops
-from cacheflow.config import CacheConfig, ModelConfig, ParallelConfig
+from vllm import cache_ops
+from vllm.config import CacheConfig, ModelConfig, ParallelConfig
 
 KVCache = Tuple[torch.Tensor, torch.Tensor]
 

cacheflow/worker/worker.py → vllm/worker/worker.py

@@ -3,16 +3,15 @@ from typing import Dict, List, Tuple
 
 import torch
 
-from cacheflow.config import (CacheConfig, ModelConfig, ParallelConfig,
+from vllm.config import (CacheConfig, ModelConfig, ParallelConfig,
                          SchedulerConfig)
-from cacheflow.model_executor import get_model, InputMetadata, set_random_seed
-from cacheflow.model_executor.parallel_utils.parallel_state import (
+from vllm.model_executor import get_model, InputMetadata, set_random_seed
+from vllm.model_executor.parallel_utils.parallel_state import (
     initialize_model_parallel, initialize_all_reduce_launcher)
-from cacheflow.sampling_params import SamplingParams
-from cacheflow.sequence import (SequenceData, SequenceGroupMetadata,
-                                SequenceOutputs)
-from cacheflow.worker.cache_engine import CacheEngine
-from cacheflow.utils import get_gpu_memory
+from vllm.sampling_params import SamplingParams
+from vllm.sequence import SequenceData, SequenceGroupMetadata, SequenceOutputs
+from vllm.worker.cache_engine import CacheEngine
+from vllm.utils import get_gpu_memory
 
 
 class Worker: