Age | Commit message (Collapse) | Author |
|
* Input 30 for QLSTM is zero point of the hidden state, ANEURALNETWORKS_INT32
* Input 31 for QLSTM is the scale of the hidden state, ANEURALNETWORKS_FLOAT32
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I1c6304fa4772895ce57286b58b2c3363c25c38e2
|
|
* Add QLSTM support for Android NN Driver
* Add overrideOutputInfo parameter to SetupAndTrackLayerOutputSlot
* Add optional condition to GetInputScalar
* Refactor Quantized 16 Bit LSTM impl
Change-Id: Ie8fa98ad5ee4a62174ef91ca80f1df62b7fde937
Signed-off-by: Keith Davis <keith.davis@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
|
|
* Read alpha parameter for ELU operation
* Created ConvertionUtils_1_3 for 1.3 Driver
* Added QAsymmS8 data type support to swizzle the tensor
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I9d66a4e8d5468efa305bb8f6c352f13d27602274
|