Skip to content

first_wave_models.md

Latest commit

聽

History

History
42 lines (39 loc) 路 9.67 KB

first_wave_models.md

File metadata and controls

42 lines (39 loc) 路 9.67 KB
Raw

The first-wave models

Model Use Case
SqueezeNet Image Classification
MobileNet Image Classification
ResNet Image Classification
TinyYOLO Object Detection
RNNoise Noise Suppression
NSNet Noise Suppression

The ops required by the first-wave models

Op SqueezeNetV1.1 ONNX TFLite MobileNetV2 ONNX TFLite ResNetV2 ONNX TFLite TinyYOLOV2 ONNX TFLite RNNoise C NSNet ONNX ONNX XLA-HLO
add 鉁旓笍 鉁旓笍 鉁旓笍 鉁旓笍 鉁旓笍 Add Add
averagePool2d 鉁旓笍 AveragePool Lowering to ReduceWindow, Add and Div.
batchNormalization (1) 鉁旓笍 鉁旓笍 鉁旓笍 BatchNormalization BatchNormInference
clamp 鉁旓笍 (6) 鉁旓笍 Clip Clamp
concat 鉁旓笍 鉁旓笍 Concat Concatenate
conv2d 鉁旓笍 鉁旓笍 (2) 鉁旓笍 鉁旓笍 Conv ConvGeneralDilated
gemm 鉁旓笍 Gemm Lowering to Broadcast, Transpose, Mul, Dot and Add
globalAveragePool (a variant of averagePool2d) 鉁旓笍 鉁旓笍 GlobalAveragePool Lowering to ReduceWindow, Add and Div.
gru 鉁旓笍 (5) 鉁旓笍 GRU Lowering to Add, Mul, and Tanh.
leakyRelu (3) 鉁旓笍 LeakyRelu Lowering to Mul, Gt and Select.
matmul 鉁旓笍 鉁旓笍 MatMul Dot
maxPool2d 鉁旓笍 鉁旓笍 鉁旓笍 MaxPool Lowering to ReduceWindow and Max.
mul 鉁旓笍 Mul Mul
relu 鉁旓笍 鉁旓笍 鉁旓笍 鉁旓笍 Relu Lowering to Max
reshape 鉁旓笍 鉁旓笍 鉁旓笍 Reshape Reshape
sigmoid 鉁旓笍 鉁旓笍 Sigmoid Lowering to Tanh
softmax (4) 鉁旓笍 鉁旓笍 鉁旓笍 Softmax Lowering to Add, Div, Exp, Max, Sub and Reduce.
split 鉁旓笍 Split Lowering to Slice
squeeze 鉁旓笍 Squeeze Lowering to Reshape
tanh 鉁旓笍 Tanh Tanh

Remarks:

  1. TFLite models don't contain batchNormalization. It is fused with other operations during the model conversion.
  2. A depthwise conv2d operation is a variant of grouped conv2d, used in MobileNet, where the groups = input_channels = output_channels.
  3. TFLite models don't contain leakyRelu. It is replaced with other operations during the model conversion.
  4. ONNX models don't contain softmax. It is implemented in the post-processing script.
  5. RNNoise implements gru as a combination of sigmoid, tanh, relu, add, split, matmul and mul.
  6. TFLite model uses Relu6 which can be lowered to clamp.