[Bugfix] Skip gs:// model paths for speculator detection (#27846)

Signed-off-by: Peter Schuurman <psch@google.com>

tests/transformers_utils/test_utils.py

@@ -0,0 +1,26 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+
+from vllm.transformers_utils.utils import is_cloud_storage, is_gcs, is_s3
+
+
+def test_is_gcs():
+    assert is_gcs("gs://model-path")
+    assert not is_gcs("s3://model-path/path-to-model")
+    assert not is_gcs("/unix/local/path")
+    assert not is_gcs("nfs://nfs-fqdn.local")
+
+
+def test_is_s3():
+    assert is_s3("s3://model-path/path-to-model")
+    assert not is_s3("gs://model-path")
+    assert not is_s3("/unix/local/path")
+    assert not is_s3("nfs://nfs-fqdn.local")
+
+
+def test_is_cloud_storage():
+    assert is_cloud_storage("gs://model-path")
+    assert is_cloud_storage("s3://model-path/path-to-model")
+    assert not is_cloud_storage("/unix/local/path")
+    assert not is_cloud_storage("nfs://nfs-fqdn.local")

vllm/engine/arg_utils.py

@@ -86,7 +86,7 @@ from vllm.transformers_utils.config import (
     is_interleaved,
     maybe_override_with_speculators,
 )
-from vllm.transformers_utils.utils import check_gguf_file, is_s3
+from vllm.transformers_utils.utils import check_gguf_file, is_cloud_storage
 from vllm.utils.argparse_utils import FlexibleArgumentParser
 from vllm.utils.mem_constants import GiB_bytes
 from vllm.utils.network_utils import get_ip
@@ -1310,10 +1310,10 @@ class EngineArgs:
 
         # Check if the model is a speculator and override model/tokenizer/config
         # BEFORE creating ModelConfig, so the config is created with the target model
-        # Skip speculator detection for S3 models since HuggingFace cannot load
+        # Skip speculator detection for cloud storage models (eg: S3, GCS) since
-        # configs directly from S3 URLs. S3 models can still use speculators with
+        # HuggingFace cannot load configs directly from S3 URLs. S3 models can still
-        # explicit --speculative-config.
+        # use speculators with explicit --speculative-config.
-        if not is_s3(self.model):
+        if not is_cloud_storage(self.model):
             (self.model, self.tokenizer, self.speculative_config) = (
                 maybe_override_with_speculators(
                     model=self.model,

vllm/transformers_utils/utils.py

@@ -19,6 +19,14 @@ def is_s3(model_or_path: str) -> bool:
     return model_or_path.lower().startswith("s3://")
 
 
+def is_gcs(model_or_path: str) -> bool:
+    return model_or_path.lower().startswith("gs://")
+
+
+def is_cloud_storage(model_or_path: str) -> bool:
+    return is_s3(model_or_path) or is_gcs(model_or_path)
+
+
 def check_gguf_file(model: str | PathLike) -> bool:
     """Check if the file is a GGUF model."""
     model = Path(model)