Intermediate Representation Suitable for INT8 Inference¶
OpenVINO Runtime CPU and GPU devices can infer models in low precision. For more details, refer to the Model Optimization Guide.
Intermediate Representation should be specifically formed to be suitable for low precision inference. Such a model is called a Low Precision IR and can be generated in two ways:
By quantizing regular IR with the Post-Training Optimization tool
Using Model Optimizer for a model pre-trained for Low Precision inference: TensorFlow pre-TFLite models (
.pbmodel file with
FakeQuantize\*operations) and ONNX quantized models. Both TensorFlow and ONNX quantized models can be prepared by Neural Network Compression Framework.
For an operation to be executed in INT8, it must have
FakeQuantize operations as inputs. For more details, see the specification of `FakeQuantize operation <doxid-openvino_docs_ops_quantization__fake_quantize_1>`.
To execute the
Convolution operation in INT8 on CPU, both data and weight inputs should have
FakeQuantize as an input operation:
Low precision IR is also suitable for FP32 and FP16 inference if a chosen plugin supports all operations of the IR. The only difference between a Low Precision IR and FP16 or FP32 IR is the existence of
FakeQuantize in the Low Precision IR. Plugins that support Low Precision Inference recognize these sub-graphs and quantize them during inference. The ones that do not, execute all operations, including
FakeQuantize, as is in the FP32 or FP16 precision.
FakeQuantize operations are present in an OpenVINO IR, it suggests to the inference device how to quantize particular operations in the model. If the device is capable, it accepts the suggestion and performs Low Precision Inference. If not, it executes the model in the floating-point precision.
Compressed Low Precision Weights¶
Weighted operations, such as
MatMul, store weights as the floating-point
Constant in the graph followed by the
FakeQuantize operation. The
Constant followed by the
FakeQuantize operation could be optimized memory-wise due to the
FakeQuantize operation semantics. The resulting weights sub-graph stores weights in Low Precision
Constant, which gets unpacked back to floating point with the
Convert operation. Weights compression replaces
FakeQuantize with optional
Multiply operation leaving output arithmetically the same and weights storing takes four times less memory.
See the visualization of
Convolution with the compressed weights:
Both Model Optimizer and Post-Training Optimization tool generate a compressed IR by default.