io loss module network optim util

dtype

Tensor

TensorOpsInterface

NATIVE_FILE gradient_functions layout

_var abs absolute add all amax amin any arange argmax argmin backward bitwiseAnd bitwiseOr bitwiseXor bytesUsed ceil clamp clip concat concatenate conv2d conv2dBackwardData cos countNonzero cumsum div elu eq erf exp eye flip floor full greaterThan greaterThanEqual gt gte hardTanh ident identity iota isinf isnan lShift leakyRelu lessThan lessThanEqual log log1p logicalAnd logicalNot logicalOr lt lte matmul maximum mean median minimum mm mod mul negate negative neq nonzero norm normalize power rShift rand randn relu relu6 reshape rint roll scalar sigmoid sign sin softmax sort sqrt std sub sum swish tanh tensor thresholdRelu tile transpose tril triu variance where wrapFLTensor