Local LRE compiled model

Category: checkpoint.lre
Version: 1.2.0
UUID: e74ff34baa480c2d590705ee108a43ebac7a5cd516389c2aef773b873b8dece8
Execution context: leip_af
Choice Priority: 10.0

Value Parameters

Name	Synonyms	Type	Values
GPU device ID to use	checkpoint.lre.device_id	scalar	int
Execution provider for ONNX runtime	checkpoint.lre.execution_provider	choice	cpu cuda tensorrt None
Path to encryption key	checkpoint.lre.key_path	scalar	path
Path to compiled model	checkpoint.lre.path, checkpoint.path	scalar	path
Encryption password	checkpoint.lre.password	scalar	string
The runtime precision for TRT models.	checkpoint.lre.precision, checkpoint.precision	choice	float32 float16 int8 None
Directory for TensorRT engine cache	checkpoint.lre.tensorrt_engine_cache	scalar	string
Directory for TensorRT timing cache	checkpoint.lre.tensorrt_timing_cache	scalar	string

Constraints

1. When using a compiled model, the dataloader batch size must match the model's batch size (train.batch_size_val == export.batch_size)

This component fits into

Name	UUID	Synonyms
Full Recipe	8b4b05...	checkpoint