Merge tag 'v0.7.2' into v0.7.2-dev

tests/compile/piecewise/test_toy_llama.py

+# SPDX-License-Identifier: Apache-2.0
 """
 Test the piecewise compilation with a simple model, comparing the output
 with and without the piecewise compilation.

tests/compile/test_basic_correctness.py

+# SPDX-License-Identifier: Apache-2.0
+
 import dataclasses
 from typing import Dict, List, Optional
 

tests/compile/test_full_graph.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pytest
 
 from vllm.config import CompilationLevel

tests/compile/test_functionalization.py

+# SPDX-License-Identifier: Apache-2.0
+
 import os
 import pytest
 import torch

tests/compile/test_fusion.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pytest
 import torch
 from compressed_tensors.quantization import FP8_DTYPE

tests/compile/test_pass_manager.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pickle
 
 import pytest

tests/compile/test_wrapper.py

+# SPDX-License-Identifier: Apache-2.0
+
 from typing import Optional
 
 import torch

tests/compile/utils.py

+# SPDX-License-Identifier: Apache-2.0
+
 import os
 
 import torch

tests/conftest.py

+# SPDX-License-Identifier: Apache-2.0
+
 import json
 import os
 import tempfile
@@ -738,6 +740,7 @@ class VllmRunner:
         images: Optional[PromptImageInput] = None,
         videos: Optional[PromptVideoInput] = None,
         audios: Optional[PromptAudioInput] = None,
+        **kwargs: Any,
     ) -> List[Tuple[List[List[int]], List[str]]]:
         inputs = self.get_inputs(prompts,
                                  images=images,
@@ -745,7 +748,8 @@ class VllmRunner:
                                  audios=audios)
 
         req_outputs = self.model.generate(inputs,
-                                          sampling_params=sampling_params)
+                                          sampling_params=sampling_params,
+                                          **kwargs)
 
         outputs: List[Tuple[List[List[int]], List[str]]] = []
         for req_output in req_outputs:
@@ -783,6 +787,7 @@ class VllmRunner:
         images: Optional[PromptImageInput] = None,
         audios: Optional[PromptAudioInput] = None,
         videos: Optional[PromptVideoInput] = None,
+        **kwargs: Any,
     ) -> Union[List[TokensTextLogprobs],
                List[TokensTextLogprobsPromptLogprobs]]:
         inputs = self.get_inputs(prompts,
@@ -791,7 +796,8 @@ class VllmRunner:
                                  audios=audios)
 
         req_outputs = self.model.generate(inputs,
-                                          sampling_params=sampling_params)
+                                          sampling_params=sampling_params,
+                                          **kwargs)
 
         toks_str_logsprobs_prompt_logprobs = (
             self._final_steps_generate_w_logprobs(req_outputs))
@@ -827,13 +833,15 @@ class VllmRunner:
         images: Optional[PromptImageInput] = None,
         videos: Optional[PromptVideoInput] = None,
         audios: Optional[PromptAudioInput] = None,
+        **kwargs: Any,
     ) -> List[Tuple[List[int], str]]:
         greedy_params = SamplingParams(temperature=0.0, max_tokens=max_tokens)
         outputs = self.generate(prompts,
                                 greedy_params,
                                 images=images,
                                 videos=videos,
-                                audios=audios)
+                                audios=audios,
+                                **kwargs)
         return [(output_ids[0], output_str[0])
                 for output_ids, output_str in outputs]
 
@@ -848,6 +856,7 @@ class VllmRunner:
         videos: Optional[PromptVideoInput] = None,
         stop_token_ids: Optional[List[int]] = None,
         stop: Optional[List[str]] = None,
+        **kwargs: Any,
     ) -> Union[List[TokensTextLogprobs],
                List[TokensTextLogprobsPromptLogprobs]]:
         greedy_logprobs_params = SamplingParams(
@@ -862,7 +871,8 @@ class VllmRunner:
                                         greedy_logprobs_params,
                                         images=images,
                                         audios=audios,
-                                        videos=videos)
+                                        videos=videos,
+                                        **kwargs)
 
     def generate_encoder_decoder_greedy_logprobs(
         self,

tests/core/block/conftest.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pytest
 
 

tests/core/block/e2e/conftest.py

+# SPDX-License-Identifier: Apache-2.0
+
 from typing import Callable, Iterable, Optional
 
 import pytest

tests/core/block/e2e/test_correctness.py

+# SPDX-License-Identifier: Apache-2.0
+
 from itertools import cycle
 
 import pytest

tests/core/block/e2e/test_correctness_sliding_window.py

+# SPDX-License-Identifier: Apache-2.0
+
 import random
 from typing import List
 

tests/core/block/test_block_manager.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pytest
 
 from vllm.core.block.utils import (STR_NOT_IMPL_ENC_DEC_PREFIX_CACHE,

tests/core/block/test_block_table.py

+# SPDX-License-Identifier: Apache-2.0
+
 from typing import List
 
 import pytest

tests/core/block/test_common.py

+# SPDX-License-Identifier: Apache-2.0
+
 import random
 
 import pytest

tests/core/block/test_cpu_gpu_block_allocator.py

+# SPDX-License-Identifier: Apache-2.0
+
 import pytest
 
 from vllm.core.block.cpu_gpu_block_allocator import CpuGpuBlockAllocator

tests/core/block/test_naive_block.py

+# SPDX-License-Identifier: Apache-2.0
+
 from typing import List, Optional
 
 import pytest

tests/core/block/test_prefix_caching_block.py

+# SPDX-License-Identifier: Apache-2.0
+
 import math
 import random
 from typing import List, Optional
@@ -63,8 +65,8 @@ class TestPrefixCachingBlock:
 
         previous_block = MagicMock(spec=PrefixCachingBlock)
         prev_block_hash = random.randint(0, 1000)
-        previous_block.content_hash = (prev_block_hash
-                                       if prev_block_has_hash else None)
+        previous_block.content_hash = (prev_block_hash if prev_block_has_hash
+                                       else hash('None'))
 
         num_to_fill = block_size if is_curr_block_full else random.randint(
             0, block_size - 1)

tests/core/test_chunked_prefill_scheduler.py

+# SPDX-License-Identifier: Apache-2.0
+
 from typing import List
 from unittest.mock import MagicMock