GemmFloat8

 Graphs in : cluster, ONNX model architecture.

A (heterogeneous) – TA : object, input tensor A. The shape of A should be (M, K) if transA is 0, or (K, M) if transA is non-zero.
B (heterogeneous) – TB : object, input tensor B. The shape of B should be (K, N) if transB is 0, or (N, K) if transB is non-zero.
C (optional, heterogeneous) – TC : object, input tensor C.
scaleA (optional, heterogeneous) – TS : object, scale of tensor A if A is float 8 tensor.
scaleB (optional, heterogeneous) – TS : object, scale of tensor B if B is float 8 tensor.
scaleY (optional, heterogeneous) – TS : object, scale of the output tensor if A or B is float 8.

 Parameters : cluster,

activation : enum, activation function, RELU or GELU or NONE.
Default value “RELU”.
 alpha : float, scalar multiplier for the product of input tensors A * B.
Default value “0”.
 beta : float, scalar multiplier for the product of input bias C.
Default value “0”.
dtype : enum, output Type. Same definition as attribute ‘to’ for operator Cast.
Default value “UNDEFINED”.
 transA : boolean, whether A should be transposed. Float 8 only supported transA=0.
Default value “False”.
 transB : boolean, whether B should be transposed. Float 8 only supported transB=1.
Default value “False”.
 training? : boolean, whether the layer is in training mode (can store data for backward).
Default value “True”.
 lda coeff : float, defines the coefficient by which the loss derivative will be multiplied before being sent to the previous layer (since during the backward run we go backwards).
Default value “1”.

 name (optional) : string, name of the node.