6.2 Release (#1751)

.gitlab-ci.yml

@@ -98,7 +98,7 @@ test_py37_pytorch:
     WHEEL_PATH: build/dist/*cp37*10_15*
 
 test_py37_tf1:
-  <<: *test_macos_pkg_with_reqs
+  <<: *test_macos_pkg
   tags:
     - macos12
   dependencies:
@@ -107,10 +107,9 @@ test_py37_tf1:
     PYTHON: "3.7"
     TEST_PACKAGE: coremltools.converters.mil.frontend.tensorflow
     WHEEL_PATH: build/dist/*cp37*10_15*
-    REQUIREMENTS: reqs/test_tf1.pip
 
 test_py37_tf2:
-  <<: *test_macos_pkg
+  <<: *test_macos_pkg_with_reqs
   tags:
     - macos12
   dependencies:
@@ -119,6 +118,7 @@ test_py37_tf2:
     PYTHON: "3.7"
     TEST_PACKAGE: coremltools.converters.mil.frontend.tensorflow2
     WHEEL_PATH: build/dist/*cp37*10_15*
+    REQUIREMENTS: reqs/test.pip
 
 test_py37_mil:
   <<: *test_macos_pkg

coremlpython/CoreMLPython.h

@@ -21,6 +21,7 @@ namespace CoreML {
         private:
             MLModel *m_model = nil;
             NSURL *compiledUrl = nil;
+            bool m_deleteCompiledModelOnExit;
         public:
             Model(const Model&) = delete;
             Model& operator=(const Model&) = delete;

coremlpython/CoreMLPython.mm

@@ -33,42 +33,60 @@ bool usingMacOS13OrHigher() {
     return (NSProtocolFromString(@"MLProgram") != nil);
 }
 
+bool isCompiledModelPath(const std::string& path) {
+    const std::string fileExtension = ".mlmodelc";
+
+    size_t start = path.length() - fileExtension.length();
+    if (path.back() == '/') {
+        start--;
+    }
+    const std::string match = path.substr(start, fileExtension.length());
+
+    return (match == fileExtension);
+}
+
 Model::~Model() {
     @autoreleasepool {
         NSFileManager *fileManager = [NSFileManager defaultManager];
-        if (compiledUrl != nil) {
+        if (compiledUrl != nil and m_deleteCompiledModelOnExit) {
             [fileManager removeItemAtURL:compiledUrl error:NULL];
         }
     }
 }
 
 Model::Model(const std::string& urlStr, const std::string& computeUnits) {
     @autoreleasepool {
-
-        // Compile the model
         NSError *error = nil;
-        NSURL *specUrl = Utils::stringToNSURL(urlStr);
-
-        // Swallow output for the very verbose coremlcompiler
-        int stdoutBack = dup(STDOUT_FILENO);
-        int devnull = open("/dev/null", O_WRONLY);
-        dup2(devnull, STDOUT_FILENO);
-
-        // Compile the model
-        compiledUrl = [MLModel compileModelAtURL:specUrl error:&error];
-
-        // Close all the file descriptors and revert back to normal
-        dup2(stdoutBack, STDOUT_FILENO);
-        close(devnull);
-        close(stdoutBack);
-
-        // Translate into a type that pybind11 can bridge to Python
-        if (error != nil) {
-            std::stringstream errmsg;
-            errmsg << "Error compiling model: \"";
-            errmsg << error.localizedDescription.UTF8String;
-            errmsg << "\".";
-            throw std::runtime_error(errmsg.str());
+
+        if (! isCompiledModelPath(urlStr)) {
+            // Compile the model
+            NSURL *specUrl = Utils::stringToNSURL(urlStr);
+
+            // Swallow output for the very verbose coremlcompiler
+            int stdoutBack = dup(STDOUT_FILENO);
+            int devnull = open("/dev/null", O_WRONLY);
+            dup2(devnull, STDOUT_FILENO);
+
+            // Compile the model
+            compiledUrl = [MLModel compileModelAtURL:specUrl error:&error];
+            m_deleteCompiledModelOnExit = true;
+
+            // Close all the file descriptors and revert back to normal
+            dup2(stdoutBack, STDOUT_FILENO);
+            close(devnull);
+            close(stdoutBack);
+
+            // Translate into a type that pybind11 can bridge to Python
+            if (error != nil) {
+                std::stringstream errmsg;
+                errmsg << "Error compiling model: \"";
+                errmsg << error.localizedDescription.UTF8String;
+                errmsg << "\".";
+                throw std::runtime_error(errmsg.str());
+            }
+        } else {
+            m_deleteCompiledModelOnExit = false;  // Don't delete user specified file
+            compiledUrl = Utils::stringToNSURL(urlStr);
         }
 
         // Set compute unit

coremltools/__init__.py

@@ -60,9 +60,6 @@
 # New versions for iOS 16.0
 _SPECIFICATION_VERSION_IOS_16 = 7
 
-# New versions for iOS 17.0
-_SPECIFICATION_VERSION_IOS_17 = 8
-
 class ComputeUnit(_Enum):
     '''
     The set of processing-unit configurations the model can use to make predictions.
@@ -79,7 +76,6 @@ class ComputeUnit(_Enum):
     _SPECIFICATION_VERSION_IOS_14: "CoreML4",
     _SPECIFICATION_VERSION_IOS_15: "CoreML5",
     _SPECIFICATION_VERSION_IOS_16: "CoreML6",
-    _SPECIFICATION_VERSION_IOS_17: "CoreML7",
 }
 
 # Default specification version for each backend

coremltools/_deps/__init__.py

@@ -147,7 +147,7 @@ def __get_sklearn_version(version):
 
 # ---------------------------------------------------------------------------------------
 _HAS_TORCH = True
-_TORCH_MAX_VERSION = "1.12.1"
+_TORCH_MAX_VERSION = "1.13.1"
 try:
     import torch
     _warn_if_above_max_supported_version("Torch", torch.__version__, _TORCH_MAX_VERSION)

coremltools/converters/mil/__init__.py

@@ -13,4 +13,4 @@
 from .input_types import (ClassifierConfig, ColorLayout, EnumeratedShapes,
                           ImageType, InputType, RangeDim, Shape, TensorType)
 from .frontend.tensorflow.tf_op_registry import register_tf_op
-from .frontend.torch import register_torch_op
+from .frontend.torch import register_torch_op

coremltools/converters/mil/_deployment_compatibility.py

@@ -8,8 +8,7 @@
 from coremltools import (_SPECIFICATION_VERSION_IOS_13,
                          _SPECIFICATION_VERSION_IOS_14,
                          _SPECIFICATION_VERSION_IOS_15,
-                         _SPECIFICATION_VERSION_IOS_16,
-                         _SPECIFICATION_VERSION_IOS_17)
+                         _SPECIFICATION_VERSION_IOS_16)
 
 
 class AvailableTarget(IntEnum):
@@ -18,7 +17,6 @@ class AvailableTarget(IntEnum):
     iOS14 = _SPECIFICATION_VERSION_IOS_14
     iOS15 = _SPECIFICATION_VERSION_IOS_15
     iOS16 = _SPECIFICATION_VERSION_IOS_16
-    iOS17 = _SPECIFICATION_VERSION_IOS_17
 
     # macOS versions (aliases of iOS versions)
     macOS15 = _SPECIFICATION_VERSION_IOS_13
@@ -28,21 +26,18 @@ class AvailableTarget(IntEnum):
     macOS11 = _SPECIFICATION_VERSION_IOS_14
     macOS12 = _SPECIFICATION_VERSION_IOS_15
     macOS13 = _SPECIFICATION_VERSION_IOS_16
-    macOS14 = _SPECIFICATION_VERSION_IOS_17
 
     # watchOS versions (aliases of iOS versions)
     watchOS6 = _SPECIFICATION_VERSION_IOS_13
     watchOS7 = _SPECIFICATION_VERSION_IOS_14
     watchOS8 = _SPECIFICATION_VERSION_IOS_15
     watchOS9 = _SPECIFICATION_VERSION_IOS_16
-    watchOS10 = _SPECIFICATION_VERSION_IOS_17
 
     # tvOS versions (aliases of iOS versions)
     tvOS13 = _SPECIFICATION_VERSION_IOS_13
     tvOS14 = _SPECIFICATION_VERSION_IOS_14
     tvOS15 = _SPECIFICATION_VERSION_IOS_15
     tvOS16 = _SPECIFICATION_VERSION_IOS_16
-    tvOS17 = _SPECIFICATION_VERSION_IOS_17
 
     # customized __str__
     def __str__(self):

coremltools/converters/mil/backend/mil/helper.py

@@ -237,9 +237,14 @@ def create_file_value_tensor(file_name, offset, dim, data_type):
 
 def types_to_proto_primitive(valuetype):
     if valuetype not in builtin_to_proto_types:
+        additional_error_msg = ""
+        if valuetype in (types.complex64, types.complex128):
+            additional_error_msg = (
+                "(MIL doesn't support complex data as model's output, please extract real and "
+                "imaginary parts explicitly.) "
+            )
         raise ValueError(
-            "Unknown type {} to map from SSA types to Proto types".format(
-                valuetype)
+            f"Unknown map from SSA type {valuetype} to Proto type. {additional_error_msg}"
         )
     return builtin_to_proto_types[valuetype]
 
@@ -302,7 +307,7 @@ def create_immediate_value(var):
         return create_tensor_value(var.val)
     elif types.is_list(var.sym_type):
         if var.elem_type == types.str:
-            return create_list_scalarvalue(var.val, np.str)
+            return create_list_scalarvalue(var.val, str)
         elif var.elem_type == types.int64:
             return create_list_scalarvalue(var.val, np.int64)
         else:

coremltools/converters/mil/backend/mil/load.py

@@ -138,9 +138,9 @@ def translate_generic_op(op, parameters, blob_writer, literal_params=[]):
 
         attr_dict["name"] = create_scalar_value(op.name)
         attr_dict["class_name"] = create_scalar_value(class_name)
-        attr_dict["input_order"] = create_list_scalarvalue(input_order, np.str)
-        attr_dict["parameters"] = create_list_scalarvalue(parameters, np.str)
-        attr_dict["weights"] = create_list_scalarvalue(weights, np.str)
+        attr_dict["input_order"] = create_list_scalarvalue(input_order, str)
+        attr_dict["parameters"] = create_list_scalarvalue(parameters, str)
+        attr_dict["weights"] = create_list_scalarvalue(weights, str)
         attr_dict["description"] = create_scalar_value(description)
 
     return pm.Operation(

coremltools/converters/mil/backend/mil/passes/test_passes.py

@@ -708,7 +708,7 @@ def prog(x):
         assert np.all(add_op.y.val == np.array([1.0, 2.0, 3.0]).reshape([1, 3, 1, 1]))
 
     @pytest.mark.parametrize(
-        "scale_type, bias_type", itertools.product([np.float, np.int32], [np.float, np.int32])
+        "scale_type, bias_type", itertools.product([np.float32, np.int32], [np.float32, np.int32])
     )
     def test_scale_bias_types(self, scale_type, bias_type):
         """

coremltools/converters/mil/backend/nn/op_mapping.py

@@ -238,10 +238,10 @@ def add_const(const_context, builder, name, val):
             return
     if not isinstance(val, (_np.ndarray, _np.generic)):
         val = _np.array([val])
-    if val.dtype != _np.float:
+    if val.dtype != float:
         # nn proto only supports float32 activation. (e.g., pred in cond op
         # needs to be converted to float)
-        val = val.astype(_np.float)
+        val = val.astype(float)
     rank = len(val.shape)
     if rank == 0:
         builder.add_load_constant_nd(
@@ -755,9 +755,9 @@ def _add_elementwise_binary(
                 builder.add_less_than(**params)
             return
 
-    if op.x.val is not None:
+    if op.x.can_be_folded_to_const():
         add_const(const_context, builder, op.x.name, op.x.val)
-    if op.y.val is not None:
+    if op.y.can_be_folded_to_const():
         if mode == "pow":
             _add_elementwise_unary(
                 const_context,

coremltools/converters/mil/converter.py

@@ -3,8 +3,6 @@
 #  Use of this source code is governed by a BSD-3-clause license that can be
 #  found in the LICENSE.txt file or at https://opensource.org/licenses/BSD-3-Clause
 
-import os as _os
-import stat as _stat
 import tempfile as _tempfile
 import warnings as _warnings
 
@@ -52,10 +50,9 @@ def __call__(self, model, *args, **kwargs):
                 msg = (
                     "Please update the minimum_deployment_target to {!s},"
                     " since op {} is only available in opset {!s} or newer."
-
                 ).format(max_opset_version, op.op_type, max_opset_version)
                 raise ValueError(msg)
-                
+
         if "inputs" in kwargs and kwargs["inputs"] is not None:
             inputs = kwargs["inputs"]
             if not isinstance(inputs, (list, tuple)):
@@ -208,11 +205,8 @@ def _mil_convert(
 
     if convert_to == 'mlprogram':
         # mil_convert_to_proto places weight files inside the weights_dir
-        weights_dir = _tempfile.mkdtemp()
-        kwargs['weights_dir'] = weights_dir
-
-        # To make sure everyone can read and write to this directory (on par with os.mkdir())
-        _os.chmod(weights_dir, _stat.S_IRWXU | _stat.S_IRWXG | _stat.S_IRWXO)
+        weights_dir = _tempfile.TemporaryDirectory()
+        kwargs["weights_dir"] = weights_dir.name
 
     proto, mil_program = mil_convert_to_proto(
                             model,
@@ -225,23 +219,28 @@ def _mil_convert(
     _reset_conversion_state()
 
     if convert_to == 'milinternal':
-        return mil_program # mil program
+        return mil_program  # mil program
     elif convert_to == 'milpython':
-        return proto # internal mil data structure
+        return proto  # internal mil data structure
 
-    elif convert_to == 'mlprogram':
-        package_path = _create_mlpackage(proto, weights_dir, kwargs.get("package_dir"))
-        return modelClass(package_path,
-                          is_temp_package=not kwargs.get('package_dir'),
-                          mil_program=mil_program,
-                          skip_model_load=kwargs.get('skip_model_load', False),
-                          compute_units=compute_units)
+    elif convert_to == "mlprogram":
+        package_path = _create_mlpackage(
+            proto, kwargs.get("weights_dir"), kwargs.get("package_dir")
+        )
+        return modelClass(
+            package_path,
+            is_temp_package=not kwargs.get("package_dir"),
+            mil_program=mil_program,
+            skip_model_load=kwargs.get("skip_model_load", False),
+            compute_units=compute_units,
+        )
 
     return modelClass(proto,
                       mil_program=mil_program,
                       skip_model_load=kwargs.get('skip_model_load', False),
                       compute_units=compute_units)
 
+
 def mil_convert_to_proto(
     model,
     convert_from,

coremltools/converters/mil/frontend/_utils.py

@@ -3,6 +3,8 @@
 #  Use of this source code is governed by a BSD-3-clause license that can be
 #  found in the LICENSE.txt file or at https://opensource.org/licenses/BSD-3-Clause
 
+from typing import List, Optional
+
 from coremltools.converters.mil.input_types import InputType
 from coremltools.converters.mil.mil import Builder as mb
 from coremltools.converters.mil.mil import Var, types
@@ -208,10 +210,10 @@ def _does_block_contain_symbolic_shape(block):
     return False
 
 
-def get_output_names(outputs):
+def get_output_names(outputs) -> Optional[List[str]]:
     """
     :param: list[ct.TensorType/ct.ImageType]
-    :return: list[str]
+    :return: list[str] or None
     """
     output_names = None
     if outputs is not None:

coremltools/converters/mil/frontend/tensorflow/converter.py

@@ -214,6 +214,10 @@ def __init__(self, tfssa, inputs=None, outputs=None, opset_version=None):
                 continue
             if any([isinstance(s, RangeDim) for s in inputtype.shape.shape]):
                 continue
+            if inputtype.name not in graph:
+                raise ValueError(
+                    f"The input {inputtype.name} provided is not in graph."
+                )
             node = graph[inputtype.name]
             shape = [-1 if is_symbolic(s) else s for s in inputtype.shape.shape]
             node.attr["_output_shapes"] = [shape]  # list of length 1