Skip to content

[RSDK-10284]: MLModelService: don't check for name if there is only 1 expected tensor, return error if expected tensors not present #394

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

Merged
merged 10 commits into from
Mar 27, 2025
Merged

[RSDK-10284]: MLModelService: don't check for name if there is only 1 expected tensor, return error if expected tensors not present #394

Changes from 2 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
c6352fe
dont do metadata check
bhaney Mar 26, 2025
3c586c1
Merge branch 'main' into make-mlmodel-server-less-restrictive
bhaney Mar 26, 2025
ad72349
makes single tensor exception, and adds error
bhaney Mar 27, 2025
117a1f1
linter
bhaney Mar 27, 2025
dd0e8c2
add type check in 1 input case too
bhaney Mar 27, 2025
c03fc15
bug fix
bhaney Mar 27, 2025
23e0930
undo the bad linting
bhaney Mar 27, 2025
439f8b2
header and comment
bhaney Mar 27, 2025
4943664
linting
bhaney Mar 27, 2025
f2cae06
small tweak
bhaney Mar 27, 2025
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
25 changes: 3 additions & 22 deletions src/viam/sdk/services/private/mlmodel_server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,29 +36,10 @@ ::grpc::Status MLModelServiceServer::Infer(
return helper.fail(::grpc::INVALID_ARGUMENT, "Called with no input tensors");
}

const auto md = mlms->metadata({});
MLModelService::named_tensor_views inputs;
for (const auto& input : md.inputs) {
const auto where = request->input_tensors().tensors().find(input.name);
if (where == request->input_tensors().tensors().end()) {
// Ignore any inputs for which we don't have metadata, since
// we can't validate the type info.
//
// TODO: Should this be an error? For now we just don't decode
// those tensors.
continue;
}
auto tensor = mlmodel::make_sdk_tensor_from_api_tensor(where->second);
const auto tensor_type = MLModelService::tensor_info::tensor_views_to_data_type(tensor);
if (tensor_type != input.data_type) {
std::ostringstream message;
using ut = std::underlying_type<MLModelService::tensor_info::data_types>::type;
message << "Tensor input `" << input.name << "` was the wrong type; expected type "
<< static_cast<ut>(input.data_type) << " but got type "
<< static_cast<ut>(tensor_type);
return helper.fail(::grpc::INVALID_ARGUMENT, message.str().c_str());
}
inputs.emplace(std::move(input.name), std::move(tensor));
for (const auto& [tensor_name, api_tensor] : request->input_tensors().tensors()) {
auto tensor = mlmodel::make_sdk_tensor_from_api_tensor(api_tensor);
inputs.emplace(std::move(tensor_name), std::move(tensor));
}

const auto outputs = mlms->infer(inputs, helper.getExtra());
Expand Down
Loading