def full_container_with_extension_class(self):
container = Container()
container.add_plugin(ContainerConfigurationResolutionPlugin())
container.add_plugin(ContainerConverterResolutionPlugin())
container.register_instance[BeanList[ConfigurationProvider]](
DictTreeConfigurationProvider({
"foo": {
"bar": {
"baz": "42",
"boo": "32"
}
},
"extension": {
"foo": {
"bar": {
"baz": "6",
"boo": "7",
"unique": "8"
}
}
}
})
)
container.register_instance[Converter[str, int]](Converter[str, int].cast())
return container
def container(self):
container = Container()
container.add_plugin(ContainerConfigurationResolutionPlugin())
container.register_instance[BeanList[ConfigurationProvider]](
DictTreeConfigurationProvider({"foo": {
"bar": {
"baz": 42
}
}}))
return container
def full_container_child(self, full_container):
full_container_child = Container(full_container)
full_container_child.register_instance[BeanList[ConfigurationProvider]](
DictTreeConfigurationProvider({
"foo": {
"bar": {
"baz": "62"
}
}
})
)
return full_container_child
def container_lacks_key_with_default(self):
container = Container()
container.add_plugin(ContainerConfigurationResolutionPlugin())
container.register_instance[BeanList[ConfigurationProvider]](
DictTreeConfigurationProvider({
"foo": {
"bar": {
"baz": "42"
}
}
})
)
container.register_instance[Converter[str, int]](Converter[str, int].cast())
return container
def main() -> None:
container = build_container([
DatasetConfig, ExportConfig, DatasetConfig, TrainingConfig,
EfficientNetConfig, AdamConfig, PreprocessingConfig
])
container.register_type[AbstractOptimizerFactory, AdamOptimizerFactory]()
container.register_type[AbstractModelFactory, EfficientNetModelFactory]()
container.register_type[AbstractPreprocessingFactory,
DefaultPreprocessingFactory]()
container.register_instance[BeanList[ConfigurationProvider]](
DictTreeConfigurationProvider({
"datasets": {
"training": {
"path": Path("/tmp/non_existant")
},
"validation": {
"path": Path("/tmp/non_existant")
}
}
}))
geffnet.config.set_exportable(True)
crop_to = container.resolve[PreprocessingConfig.crop_to]()
export_config: ExportConfig = container.resolve[ExportConfig]()
kwargs = container.get_kwargs_to_inject(Ori3nt8LightningModel)
model = Ori3nt8LightningModel.load_from_checkpoint(checkpoint_path=str(
export_config.checkpoint_path),
**kwargs)
model.freeze()
x = torch.normal(0, 1, size=(4, 3, crop_to, crop_to)).to(dtype=torch.float)
outputs = model(x)
input_names = ["x"]
outputs = model(x)
traced = torch.jit.trace(model, x)
buf = BytesIO()
torch.jit.save(traced, buf)
buf.seek(0)
model = torch.jit.load(buf)
buf = BytesIO()
torch.onnx.export(
model,
x,
buf,
input_names=input_names,
example_outputs=outputs,
operator_export_type=torch.onnx.OperatorExportTypes.ONNX_ATEN_FALLBACK,
opset_version=12,
do_constant_folding=True)
buf.seek(0)
onnx_model = onnx.load(buf)
# Set of optimization passes used in gen-efficientnet-pytorch:
# https://github.com/rwightman/gen-efficientnet-pytorch/blob/master/onnx_optimize.py#L37
passes = [
"eliminate_identity", "eliminate_nop_dropout", "eliminate_nop_pad",
"eliminate_nop_transpose", "eliminate_unused_initializer",
"extract_constant_to_initializer", "fuse_add_bias_into_conv",
"fuse_bn_into_conv", "fuse_consecutive_concats",
"fuse_consecutive_reduce_unsqueeze", "fuse_consecutive_squeezes",
"fuse_consecutive_transposes", "fuse_pad_into_conv"
]
optimized_model = optimizer.optimize(onnx_model, passes)
onnx.save(optimized_model, str(export_config.output_path))
def test_dict_tree_provider(self):
provider = DictTreeConfigurationProvider({})
provider.set_value(ExampleConfigurationClass.property, 42)
assert provider.get_value(["foo", "bar", "baz"]) == 42