Using these changes and

```shell
TORCH_MLIR_ENABLE_LTC=0 TORCH_MLIR_ENABLE_JIT_IR_IMPORTER=0 CMAKE_GENERATOR=Ninja python setup.py bdist_wheel --dist-dir wheelhouse
```

I can build a wheel using which PI will successfully compile/lower e2e model examples.

python/torch_mlir/__init__.py

@@ -2,19 +2,26 @@
 # See https://llvm.org/LICENSE.txt for license information.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 # Also available under a BSD-style license. See LICENSE.
-
+import warnings
 from typing import Optional, Sequence, Union, List, Dict, Tuple
 from enum import Enum
 
 import sys
 from io import StringIO
 
-from torch._functorch.compile_utils import strip_overloads
-import torch
+try:
+    from torch._functorch.compile_utils import strip_overloads
+    import torch
+except ImportError:
+    warnings.warn("PyTorch not installed")
 
 from torch_mlir.passmanager import PassManager
 from .compiler_utils import run_pipeline_with_repro_report
-from torch_mlir.dialects.torch.importer.jit_ir import ClassAnnotator, ImportOptions, ModuleBuilder
+from torch_mlir.dialects import torch as torch_dialect
+try:
+    from torch_dialect.importer.jit_ir import ClassAnnotator, ImportOptions, ModuleBuilder
+except ImportError:
+    warnings.warn("torch-mlir JIT IR Importer not installed")
 
 
 class OutputType(Enum):
@@ -88,7 +95,7 @@ class TensorPlaceholder:
     ```
     """
 
-    def __init__(self, shape: List[int], dtype: torch.dtype):
+    def __init__(self, shape: List[int], dtype: "torch.dtype"):
         """Create a tensor with shape `shape` and dtype `dtype`.
 
         Args:
@@ -100,7 +107,7 @@ def __init__(self, shape: List[int], dtype: torch.dtype):
         self.dtype = dtype
 
     @staticmethod
-    def like(tensor: torch.Tensor, dynamic_axes: List[int] = None):
+    def like(tensor: "torch.Tensor", dynamic_axes: List[int] = None):
         """Create a tensor placeholder that is like the given tensor.
 
         Args:
@@ -119,7 +126,7 @@ def like(tensor: torch.Tensor, dynamic_axes: List[int] = None):
         return TensorPlaceholder(shape, tensor.dtype)
 
 
-_example_arg = Union[TensorPlaceholder, torch.Tensor]
+_example_arg = Union[TensorPlaceholder, "torch.Tensor"]
 _example_args_for_one_method = Union[_example_arg, Sequence[_example_arg]]
 _example_args = Union[_example_args_for_one_method, "ExampleArgs"]
 
@@ -246,7 +253,7 @@ def _get_for_tracing(
 }
 
 
-def compile(model: torch.nn.Module,
+def compile(model: "torch.nn.Module",
             example_args: _example_args,
             output_type: Union[str, "OutputType"] = OutputType.TORCH,
             use_tracing: bool = False,

setup.py

@@ -45,8 +45,9 @@
 
 PACKAGE_VERSION = os.environ.get("TORCH_MLIR_PYTHON_PACKAGE_VERSION") or "0.0.1"
 
-# If true, enable LTC build by default
+# If true, enable LTC and JIT IR importer build by default
 TORCH_MLIR_ENABLE_LTC_DEFAULT = True
+TORCH_MLIR_ENABLE_JIT_IR_IMPORTER = int(os.environ.get('TORCH_MLIR_ENABLE_JIT_IR_IMPORTER', True))
 
 # Build phase discovery is unreliable. Just tell it what phases to run.
 class CustomBuild(_build):
@@ -90,6 +91,7 @@ def run(self):
                 f"-DCMAKE_C_VISIBILITY_PRESET=hidden",
                 f"-DCMAKE_CXX_VISIBILITY_PRESET=hidden",
                 f"-DTORCH_MLIR_ENABLE_LTC={'ON' if enable_ltc else 'OFF'}",
+                f"-DTORCH_MLIR_ENABLE_JIT_IR_IMPORTER={'ON' if TORCH_MLIR_ENABLE_JIT_IR_IMPORTER else 'OFF'}",
             ]
 
             os.makedirs(cmake_build_dir, exist_ok=True)
@@ -159,9 +161,7 @@ def build_extension(self, ext):
     ext_modules=[
         CMakeExtension("torch_mlir._mlir_libs._jit_ir_importer"),
     ],
-    install_requires=[
-        "numpy",
-        f"torch=={torch.__version__}".split("+", 1)[0],
-    ],
+    install_requires=["numpy", ] + [
+        f"torch=={torch.__version__}".split("+", 1)[0], ] if TORCH_MLIR_ENABLE_JIT_IR_IMPORTER else [],
     zip_safe=False,
 )