saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-23 19:32:23 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Allow saving of large models after optimization (github issue 12882) (#16440)

Browse source 
### Description
Allow saving of large models after optimization.

### Motivation and Context
Addresses https://github.com/microsoft/onnxruntime/issues/12882
This commit is contained in:
Pranav Sharma 2023-06-21 22:46:26 -07:00 committed by GitHub
parent 89f8f20a61
commit a270d8407e
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 50 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
include/onnxruntime/core/session/onnxruntime_session_options_config_keys.h
View file

@ -216,4 +216,12 @@ static const char* const kDebugLayoutTransformation = "session.debug_layout_tran
// Option values:
// - "0": CPU EP fallback is not disabled. [DEFAULT]
// - "1": CPU EP fallback is disabled.
static const char* const kOrtSessionOptionsDisableCPUEPFallback = "session.disable_cpu_ep_fallback";
static const char* const kOrtSessionOptionsDisableCPUEPFallback = "session.disable_cpu_ep_fallback";
// Use this config when serializing a large model after optimization to specify an external initializers file
static const char* const kOrtSessionOptionsOptimizedModelExternalInitializersFileName =
"session.optimized_model_external_initializers_file_name";
// Use this config to control the minimum size of the initializer when externalizing it during serialization
static const char* const kOrtSessionOptionsOptimizedModelExternalInitializersMinSizeInBytes =
"session.optimized_model_external_initializers_min_size_in_bytes";

15
onnxruntime/core/session/inference_session.cc
View file

@ -1725,7 +1725,20 @@ common::Status InferenceSession::Initialize() {
if (saving_ort_format) {
ORT_RETURN_IF_ERROR_SESSIONID_(SaveToOrtFormat(session_options_.optimized_model_filepath));
} else {
ORT_RETURN_IF_ERROR_SESSIONID_(Model::Save(*model_, session_options_.optimized_model_filepath));
const std::string optimized_model_external_initializers_file_name =
session_options_.config_options.GetConfigOrDefault(
kOrtSessionOptionsOptimizedModelExternalInitializersFileName, "");
if (optimized_model_external_initializers_file_name.empty()) {
ORT_RETURN_IF_ERROR_SESSIONID_(Model::Save(*model_, session_options_.optimized_model_filepath));
} else {
const size_t optimized_model_external_initializers_min_size_in_bytes =
ParseStringWithClassicLocale<size_t>(session_options_.config_options.GetConfigOrDefault(
kOrtSessionOptionsOptimizedModelExternalInitializersMinSizeInBytes, "1024"));
ORT_RETURN_IF_ERROR_SESSIONID_(Model::SaveWithExternalInitializers(*model_,
session_options_.optimized_model_filepath,
optimized_model_external_initializers_file_name,
optimized_model_external_initializers_min_size_in_bytes));
}
}
}

27
onnxruntime/test/python/onnxruntime_test_python.py
View file

@ -94,6 +94,33 @@ class TestInferenceSession(unittest.TestCase):
else:
raise onnxruntime_error
def testModelSerializationWithExternalInitializers(self): # noqa: N802
try:
so = onnxrt.SessionOptions()
so.log_severity_level = 1
so.logid = "TestModelSerializationWithExternalInitializers"
so.optimized_model_filepath = "./model_with_external_initializers.onnx"
external_initializers_file = "external_initializers.bin"
so.add_session_config_entry(
"session.optimized_model_external_initializers_file_name", external_initializers_file
)
so.add_session_config_entry("session.optimized_model_external_initializers_min_size_in_bytes", "100")
onnxrt.InferenceSession(
get_name("mnist.onnx"),
sess_options=so,
providers=["CPUExecutionProvider"],
)
self.assertTrue(os.path.isfile(so.optimized_model_filepath))
self.assertTrue(os.path.isfile(external_initializers_file))
except Fail as onnxruntime_error:
if (
str(onnxruntime_error) == "[ONNXRuntimeError] : 1 : FAIL : Unable to serialize model as it contains"
" compiled nodes. Please disable any execution providers which generate compiled nodes."
):
pass
else:
raise onnxruntime_error
def testGetProviders(self): # noqa: N802
self.assertTrue("CPUExecutionProvider" in onnxrt.get_available_providers())
# get_all_providers() returns the default EP order from highest to lowest.