[ci/lint] Add back default arg for pre-commit (#12279)

Signed-off-by: kevin <[email protected]>

Author: khluu

.github/workflows/pre-commit.yml

@@ -16,4 +16,4 @@ jobs:
     - run: echo "::add-matcher::.github/workflows/matchers/actionlint.json"
     - uses: pre-commit/action@2c7b3805fd2a0fd8c1884dcaebf91fc102a13ecd # v3.0.1
       with:
-        extra_args: --hook-stage manual
+        extra_args: --all-files --hook-stage manual

tests/models/decoder_only/language/test_gguf.py

@@ -74,11 +74,7 @@ def gguf_model(self):
 )
 
 MODELS = [
-    LLAMA_CONFIG,
-    QWEN2_CONFIG,
-    PHI3_CONFIG,
-    GPT2_CONFIG,
-    STABLELM_CONFIG,
+    LLAMA_CONFIG, QWEN2_CONFIG, PHI3_CONFIG, GPT2_CONFIG, STABLELM_CONFIG,
     DOLPHIN_CONFIG
     # STARCODER_CONFIG, # broken
 ]
@@ -114,11 +110,12 @@ def test_models(
             messages, tokenize=False, add_generation_prompt=True)
 
     # Run unquantized model.
-    with vllm_runner(model_name=model.original_model,
-                     enforce_eager=True, # faster tests
-                     dtype=dtype,
-                     max_model_len=MAX_MODEL_LEN,
-                     tensor_parallel_size=tp_size) as original_model:
+    with vllm_runner(
+            model_name=model.original_model,
+            enforce_eager=True,  # faster tests
+            dtype=dtype,
+            max_model_len=MAX_MODEL_LEN,
+            tensor_parallel_size=tp_size) as original_model:
         original_outputs = original_model.generate_greedy_logprobs(
             example_prompts[:-1], max_tokens, num_logprobs)
 

vllm/model_executor/models/paligemma.py

@@ -147,7 +147,7 @@ class PaliGemmaForConditionalGeneration(nn.Module, SupportsMultiModal,
             "up_proj",
         ],
     }
-    
+
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
         config = vllm_config.model_config.hf_config

vllm/model_executor/models/siglip.py

@@ -348,12 +348,10 @@ def __init__(
         if quant_config and quant_config.get_name() == "bitsandbytes":
             quantizable = True
         else:
-            # For other quantization, we require the hidden size to be a 
+            # For other quantization, we require the hidden size to be a
             # multiple of 64
-            quantizable = (
-                config.hidden_size % 64 == 0
-                and config.intermediate_size % 64 == 0
-            )
+            quantizable = (config.hidden_size % 64 == 0
+                           and config.intermediate_size % 64 == 0)
         self.fc1 = ColumnParallelLinear(
             config.hidden_size,
             config.intermediate_size,

vllm/platforms/__init__.py

@@ -101,7 +101,7 @@ def cpu_platform_plugin() -> Optional[str]:
     try:
         from importlib.metadata import version
         is_cpu = "cpu" in version("vllm")
-        if is_cpu == False:
+        if not is_cpu:
             import platform
             is_cpu = platform.machine().lower().startswith("arm")
 

vllm/v1/stats/common.py

@@ -10,10 +10,11 @@
 from vllm.sampling_params import SamplingParams
 
 
-class RequestStatsUpdate(msgspec.Struct,
-                         array_like=True,
-                         omit_defaults=True,
-                         gc=False):
+class RequestStatsUpdate(
+        msgspec.Struct,  # type: ignore
+        array_like=True,
+        omit_defaults=True,
+        gc=False):
     """
     An update to the request stats.
 
@@ -341,16 +342,16 @@ def update_from(self, update: "RequestStatsUpdate"):
             self.queued_ts_s = ts
         elif update.type == RequestStatsUpdate.Type.PREFILLING:
             self.prefill_start_ts_s_lst.append(ts)
-            self.num_cached_tokens = update.num_cached_tokens
-            self.num_computed_tokens = update.num_computed_tokens
+            self.num_cached_tokens = update.num_cached_tokens or 0
+            self.num_computed_tokens = update.num_computed_tokens or 0
         elif update.type == RequestStatsUpdate.Type.PREEMPTED:
             self._reset_for_preemption(ts)
         elif update.type == RequestStatsUpdate.Type.DECODING:
             self.decoding_ts_s_lst.append(ts)
         elif update.type == RequestStatsUpdate.Type.DETOKENIZED:
             self._record_detokenized_output(
                 ts,
-                update.num_new_tokens,
+                update.num_new_tokens or 0,
             )
         elif update.type == RequestStatsUpdate.Type.FINISHED:
             self.finished_ts_s = ts
@@ -425,10 +426,11 @@ class EngineCoreProcessStats:
     output_queue_size: Optional[int] = None
 
 
-class EngineCoreStatsSnapshot(msgspec.Struct,
-                              array_like=True,
-                              omit_defaults=True,
-                              gc=False):
+class EngineCoreStatsSnapshot(
+        msgspec.Struct,  # type: ignore
+        array_like=True,
+        omit_defaults=True,
+        gc=False):
     """
     A snapshot of the EngineCore's current stats over a period of time.
     """