Provenance and compatibility metadata for a model artifact.
More...
#include <gpu_neural_model.hpp>
Provenance and compatibility metadata for a model artifact.
String views are copied into the model.
◆ architecture
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::architecture {} |
Free-form, except that a Tcn model whose architecture is "nam.tcn" carries its own topology and needs no shape.
◆ artifact_id
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::artifact_id {} |
◆ artifact_sha256
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::artifact_sha256 {} |
Lower- or upper-case hex SHA-256 of the artifact.
◆ artifact_size_bytes
| std::uint64_t pulp::gpu_audio::GpuNeuralModelManifest::artifact_size_bytes = 0 |
◆ license
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::license {} |
◆ local_research
| bool pulp::gpu_audio::GpuNeuralModelManifest::local_research = false |
◆ model_id
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::model_id {} |
◆ model_version
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::model_version {} |
◆ redistributable
| bool pulp::gpu_audio::GpuNeuralModelManifest::redistributable = false |
A shipped model must be redistributable unless built for local research.
◆ runtime
| std::string_view pulp::gpu_audio::GpuNeuralModelManifest::runtime {} |
◆ sample_rate_hz
| std::uint32_t pulp::gpu_audio::GpuNeuralModelManifest::sample_rate_hz = 0 |
The documentation for this struct was generated from the following file: