Skip to content

Internal Change Only #2711

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
copybara-service wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Internal Change Only #2711

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion runtime/executor/BUILD
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -386,6 +386,7 @@ cc_library(
"@litert//litert/cc:litert_model",
"@litert//litert/cc:litert_ranked_tensor_type",
"//runtime/components:model_resources",
"//runtime/proto:llm_model_type_cc_proto",
"//runtime/util:litert_status_util",
] + select({
"@platforms//os:ios": [],
Expand All @@ -404,6 +405,7 @@ cc_test(
":llm_executor_settings",
":llm_litert_compiled_model_executor_factory",
"@com_google_googletest//:gtest_main",
"@com_google_absl//absl/status",
"@com_google_absl//absl/status:statusor",
"@com_google_absl//absl/strings",
"@litert//litert/cc:litert_environment",
Expand All @@ -412,7 +414,6 @@ cc_test(
"//runtime/components:model_resources_litert_lm",
"//runtime/util:litert_lm_loader",
"//runtime/util:litert_status_util",
"//runtime/util:model_asset_bundle_resources",
"//runtime/util:scoped_file",
"//runtime/util:test_utils",
],
Expand Down
4 changes: 3 additions & 1 deletion runtime/executor/llm_litert_compiled_model_executor_factory.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,6 +34,7 @@
#include "runtime/executor/llm_executor.h"
#include "runtime/executor/llm_executor_settings.h"
#include "runtime/executor/llm_litert_compiled_model_executor.h"
#include "runtime/proto/llm_model_type.pb.h"
#include "runtime/util/status_macros.h"

#if !defined(LITERT_DISABLE_NPU)
Expand Down Expand Up @@ -131,10 +132,11 @@ absl::StatusOr<std::unique_ptr<LlmExecutor>>
CreateCpuOrGpuLlmLiteRtCompiledModelExecutor(
LlmExecutorSettings executor_settings, Environment& lrt_env,
ModelResources& resources) {
std::unique_ptr<LlmExecutor> executor;

ASSIGN_OR_RETURN(const litert::Model* litert_model,
resources.GetTFLiteModel(ModelType::kTfLitePrefillDecode));

std::unique_ptr<LlmExecutor> executor;
ASSIGN_OR_RETURN(bool is_dynamic_model, IsDynamicModel(*litert_model));
if (is_dynamic_model) {
ASSIGN_OR_RETURN(executor, LlmLiteRtCompiledModelExecutorDynamic::Create(
Expand Down
2 changes: 1 addition & 1 deletion runtime/executor/llm_litert_compiled_model_executor_factory_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@

#include <gmock/gmock.h>
#include <gtest/gtest.h>
#include "absl/status/status.h" // from @com_google_absl
#include "absl/status/statusor.h" // from @com_google_absl
#include "absl/strings/string_view.h" // from @com_google_absl
#include "litert/cc/litert_environment.h" // from @litert
Expand All @@ -31,7 +32,6 @@
#include "runtime/executor/llm_executor_settings.h"
#include "runtime/util/litert_lm_loader.h"
#include "runtime/util/scoped_file.h"
#include "runtime/util/status_macros.h"
#include "runtime/util/test_utils.h" // IWYU pragma: keep

namespace litert::lm {
Expand Down
2 changes: 1 addition & 1 deletion runtime/framework/resource_management/serial_execution_manager.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -786,7 +786,7 @@ absl::Status SerialExecutionManager::AddDecodeTask(
session_info->stop_token_detector->ResetBatch(num_output_candidates);
std::optional<Sampler*> optional_sampler = std::nullopt;
std::optional<litert::TensorBuffer> decoded_ids_buffer = std::nullopt;
if (session_info->sampler != nullptr) {
if (session_info->sampler != nullptr && !is_text_diffusion) {
optional_sampler = session_info->sampler.get();
std::vector<int> decoded_ids(num_output_candidates,
session_info->last_prefill_token_id);
Expand Down
3 changes: 2 additions & 1 deletion runtime/framework/resource_management/threaded_execution_manager.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -908,7 +908,8 @@ absl::Status ThreadedExecutionManager::AddDecodeTask(
session_info->stop_token_detector->ResetBatch(num_output_candidates);
std::optional<Sampler*> optional_sampler = std::nullopt;
std::optional<litert::TensorBuffer> decoded_ids_buffer = std::nullopt;
if (session_info->sampler != nullptr) {
bool is_text_diffusion = false;
if (session_info->sampler != nullptr && !is_text_diffusion) {
optional_sampler = session_info->sampler.get();
std::vector<int> decoded_ids(num_output_candidates,
session_info->last_prefill_token_id);
Expand Down
Loading