Module: onnx_inference_workflow
Workflow class for onnx inference workflows.
This class is responsible for loading & running an onnx model.
Models can be loaded in two ways:
-
Preloading: The model is loaded & session is started in the setup method. This happens in the
setup()
method if model source and load args are provided when the class is instantiated. -
On-demand: The model is loaded with an inference request. This happens if model source and load args are provided with the input (see the optional fields in the
ONNXInferenceInput
class).
Loaded models are cached in-memory using an LRU cache. The cache size can be configured
using the ONNX_MODEL_LRU_CACHE_SIZE
environment variable.
Additional Installations
Since this workflow uses some additional libraries, you'll need to install
infernet-ml[onnx_inference]
. Alternatively, you can install those packages directly.
The optional dependencies "[onnx_inference]"
are provided for your
convenience.
Example Usage
from infernet_ml.utils.common_types import TensorInput
from infernet_ml.utils.model_loader import ModelSource, HFLoadArgs
from infernet_ml.workflows.inference.onnx_inference_workflow import (
ONNXInferenceInput,
ONNXInferenceWorkflow,
)
def main():
input_data = ONNXInferenceInput(
inputs={
"input": TensorInput(
values=[[1.0380048, 0.5586108, 1.1037828, 1.712096]],
shape=(1, 4),
dtype="float",
)
},
model_source=ModelSource.HUGGINGFACE_HUB,
load_args=HFLoadArgs(
repo_id="Ritual-Net/iris-classification",
filename="iris.onnx",
),
)
workflow = ONNXInferenceWorkflow().setup()
result = workflow.inference(input_data)
print(result)
if __name__ == "__main__":
main()
Outputs:
[TensorOutput(values=array([0.00101515, 0.01439102, 0.98459375], dtype=float32), dtype='float32', shape=(1, 3))]
Input Format
Input format is an instance of the ONNXInferenceInput
class. The fields are:
inputs
: Dict[str,TensorInput
]: Each key corresponds to an input tensor name.model_source
: Optional[ModelSource
]: Source of the model to be loadedload_args
: Optional[LoadArgs]: Arguments to be passed to the model loader, optiosn are
ONNXInferenceInput
Bases: BaseModel
Input data for ONNX inference workflows. If model source and load args are provided, the model is loaded & session is started. Otherwise, if the class is instantiated with a model source and load args, the model is preloaded in the setup method.
Input Format
Input format is a dictionary of input tensors. Each key corresponds to the name of
the input nodes defined in the onnx model. The values are of type TensorInput
.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
inputs |
Dict[str, TensorInput]: Each key corresponds to an input tensor name. |
required | |
model_source |
Optional[ModelSource]: Source of the model to be loaded |
required | |
load_args |
Optional[LoadArgs]: Arguments to be passed to the model loader |
required |
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
ONNXInferenceWorkflow
Bases: BaseInferenceWorkflow
Inference workflow for ONNX models.
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 |
|
__init__(model_source=None, load_args=None, *args, **kwargs)
Parameters:
Name | Type | Description | Default |
---|---|---|---|
model_source |
Optional[ModelSource]
|
Optional[ModelSource]: Source of the model to be loaded |
None
|
load_args |
Optional[LoadArgs]
|
Optional[LoadArgs]: Arguments to be passed to the model loader |
None
|
*args |
Any
|
|
()
|
**kwargs |
Any
|
|
{}
|
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
do_preprocessing(input_data)
Convert the input data to a dictionary of torch tensors.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
input_data |
ONNXInferenceInput
|
|
required |
Returns:
Type | Description |
---|---|
InferenceSession
|
Dict[str, torch.Tensor]: Dictionary of input tensors. Keys are the model |
Dict[str, Tensor]
|
input node names. |
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
do_run_model(_input)
Run the model with the input data.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
_input |
Tuple[InferenceSession, Dict[str, Tensor]]
|
Tuple[InferenceSession, Dict[str, torch.Tensor]]: Tuple containing |
required |
Returns:
Name | Type | Description |
---|---|---|
ONNXInferenceResult |
ONNXInferenceResult
|
List of output tensors from the model |
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
do_setup()
If model source and load args are provided, preloads the model & starts the session. Otherwise, does nothing & model is loaded with an inference request.
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
do_stream(preprocessed_input)
Streaming inference is not supported for ONNX models.
get_session(model_source, load_args)
Load the model and start the inference session.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
model_source |
ModelSource
|
|
required |
load_args |
LoadArgs
|
|
required |
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
inference(input_data)
Inference method for the workflow. Overridden to add type hints.
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
setup()
Setup method for the workflow. Overridden to add type hints.
TensorOutput
Bases: BaseModel
Output tensor from the model.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
values |
|
required | |
dtype |
|
required | |
shape |
Tuple[int, ...]: Shape of the tensor |
required |
Source code in src/infernet_ml/workflows/inference/onnx_inference_workflow.py
load_model_and_start_session(model_source, load_args)
cached
Load the model and start the inference session.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
model_source |
ModelSource
|
|
required |
load_args |
LoadArgs
|
|
required |
Returns:
Name | Type | Description |
---|---|---|
InferenceSession |
InferenceSession
|
Inference session for the model |