vm.onnx

Operators

Name Supported Reference
Abs v dragon.math.abs()
Acos    
Acosh    
Add v dragon.math.add()
And v dragon.bitwise.bitwise_and()
ArgMax v dragon.math.argmax()
ArgMin v dragon.math.argmin()
Asin    
Asinh    
Atan    
Atanh    
AveragePool v dragon.nn.pool2d()
BatchNormalization v dragon.nn.batch_norm()
BitShift    
Cast v dragon.cast()
Ceil v dragon.math.ceil()
Clip v dragon.math.clip()
Compress    
Concat v dragon.concat()
ConcatFromSequence    
Constant    
ConstantOfShape    
Conv v dragon.nn.conv2d()
ConvInteger    
ConvTranspose v dragon.nn.conv2d_transpose()
Cos v dragon.math.cos()
Cosh    
CumSum v dragon.math.cumsum()
DepthToSpace v dragon.nn.depth_to_space()
DequantizeLinear    
Det    
Div v dragon.math.div()
Dropout v dragon.nn.dropout()
Einsum    
Elu v dragon.nn.elu()
Equal v dragon.math.equal()
Erf    
Exp v dragon.math.exp()
Expand v dragon.broadcast_to()
EyeLike v dragon.eye_like()
Flatten v dragon.flatten()
Floor v dragon.math.floor()
GRU v dragon.nn.GRU()
Gather v dragon.index_select()
GatherElements    
GatherND    
Gemm v dragon.nn.fully_connected()
GlobalAveragePool v dragon.nn.pool2d()
GlobalLpPool    
GlobalMaxPool v dragon.nn.pool2d()
Greater v dragon.math.greater()
HardSigmoid    
Hardmax    
Identity    
If    
InstanceNormalization v dragon.nn.instance_norm()
IsInf v dragon.math.is_inf()
IsNaN v dragon.math.is_nan()
LRN v dragon.nn.lrn()
LSTM v dragon.nn.LSTM()
LeakyRelu v dragon.nn.leaky_relu()
Less v dragon.math.less()
Log v dragon.math.log()
LogSoftmax v dragon.nn.log_softmax()
Loop    
LpNormalization v dragon.math.lp_normalize()
LpPool    
MatMul v dragon.math.matmul()
MatMulInteger    
Max v dragon.math.maximum()
MaxPool v dragon.nn.pool2d()
MaxRoiPool v dragon.vision.roi_pool()
MaxUnpool    
Mean v dragon.math.add()
Min v dragon.math.minimum()
Mod    
Mul v dragon.math.mul()
Multinomial v dragon.random.multinomial()
Neg v dragon.math.negative()
NonMaxSuppression    
NonZero v dragon.nonzero()
Not v dragon.bitwise.invert()
OneHot v dragon.one_hot()
Or v dragon.bitwise.bitwise_or()
PRelu v dragon.nn.prelu()
Pad v dragon.pad()
Pow v dragon.math.pow()
QLinearConv    
QLinearMatMul    
QuantizeLinear    
RNN v dragon.nn.RNN()
RandomNormal v dragon.random.normal()
RandomNormalLike v dragon.random.normal_like()
RandomUniform v dragon.random.uniform()
RandomUniformLike v dragon.random.uniform_like()
Reciprocal v dragon.math.reciprocal()
ReduceL1    
ReduceL2    
ReduceLogSum    
ReduceLogSumExp    
ReduceMax v dragon.math.max()
ReduceMean v dragon.math.mean()
ReduceMin v dragon.math.min()
ReduceProd    
ReduceSum v dragon.math.sum()
ReduceSumSquare    
Relu v dragon.nn.relu()
Reshape v dragon.reshape()
Resize v dragon.vision.resize()
ReverseSequence    
RoiAlign v dragon.vision.roi_align()
Round v dragon.math.round()
Scan    
Scatter    
ScatterElements    
ScatterND    
Selu v dragon.nn.selu()
SequenceAt    
SequenceConstruct    
SequenceEmpty    
SequenceErase    
SequenceInsert    
SequenceLength    
Shape v dragon.shape()
Shrink    
Sigmoid v dragon.math.sigmoid()
Sign v dragon.math.sign()
Sin v dragon.math.sin()
Sinh    
Size    
Slice v dragon.slice()
Softmax v dragon.nn.softmax()
Softplus    
Softsign    
SpaceToDepth v dragon.nn.space_to_depth()
Split v dragon.split()
SplitToSequence    
Sqrt v dragon.math.sqrt()
Squeeze v dragon.squeeze()
StringNormalizer    
Sub v dragon.math.sub()
Sum v dragon.math.add()
Tan    
Tanh v dragon.math.tanh()
TfIdfVectorizer    
ThresholdedRelu    
Tile v dragon.tile()
TopK v dragon.math.top_k()
Transpose v dragon.transpose()
Unique v dragon.unique()
Unsqueeze v dragon.unsqueeze()
Upsample v dragon.vision.resize()
Where v dragon.where()
Xor v dragon.bitwise.bitwise_xor()

Classes

class Shell : Context-manger to export or load onnx models.