Module fully_connected 

Fully Connected (Linear / Dense) and Batch Matrix Multiplication operations.

Functions

batch_matmul_s8

Performs Batch Matrix Multiplication (BatchMatMul) for int8 tensors.

batch_matmul_s16

Performs Batch Matrix Multiplication (BatchMatMul) for int16 tensors.

fully_connected_per_channel_s8

Performs per-channel quantized int8 Fully Connected layer.

fully_connected_s8

Performs per-tensor quantized int8 Fully Connected layer.

fully_connected_s16

Performs int16 Fully Connected layer.