Fix CI (#9012)

python/sglang/srt/layers/quantization/__init__.py

@@ -26,8 +26,9 @@ try:
     from vllm.model_executor.layers.quantization.tpu_int8 import Int8TpuConfig
 
     VLLM_AVAILABLE = True
-except ImportError:
+except ImportError as e:
     VLLM_AVAILABLE = False
+    VLLM_IMPORT_ERROR = e
 
     # Define empty classes as placeholders when vllm is not available
     class DummyConfig:
@@ -137,7 +138,8 @@ def get_quantization_config(quantization: str) -> Type[QuantizationConfig]:
     if quantization in VLLM_QUANTIZATION_METHODS and not VLLM_AVAILABLE:
         raise ValueError(
             f"{quantization} quantization requires some operators from vllm. "
-            "Please install vllm by `pip install vllm==0.9.0.1`"
+            f"Please install vllm by `pip install vllm==0.9.0.1`\n"
+            f"Import error: {VLLM_IMPORT_ERROR}"
         )
 
     return QUANTIZATION_METHODS[quantization]

python/sglang/srt/layers/quantization/modelopt_quant.py

@@ -1,9 +1,8 @@
 # Adapted from https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/layers/quantization/modelopt.py
 from __future__ import annotations
 
-import importlib.util
 import logging
-from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Union
+from typing import TYPE_CHECKING, Any, Dict, List, Optional
 
 import torch
 from torch.nn.parameter import Parameter
@@ -42,11 +41,7 @@ if is_cuda():
 
 try:
     from flashinfer import mm_fp4 as fp4_gemm
-    from flashinfer import (
-        reorder_rows_for_gated_act_gemm,
-        shuffle_matrix_a,
-        shuffle_matrix_sf_a,
-    )
+    from flashinfer import reorder_rows_for_gated_act_gemm, shuffle_matrix_sf_a
 
     enable_flashinfer_fp4_gemm = True
 except ImportError: