def execute(cls, params, in_tensors, qrec: QuantizationRecordBase,
**kwargs):
in_tensors = qrec.prepare_inputs(params, in_tensors, ktype="symmetric")
if params.transpose_in:
in_tensors = [(np.transpose(in_tensor, params.transpose_in[idx])
if params.transpose_in[idx] else in_tensor)
for idx, in_tensor in enumerate(in_tensors)]
if isinstance(params, Broadcastable) and params.is_broadcasted:
in_tensors = params.broadcast_inputs(in_tensors)
func = PIECEWISE_OPS[params.__class__]
op = func['op']
if func['is_mult']:
qrec.set_scale(in_idx=(0, 1), out_idx=0)
i1 = in_tensors[0].astype(np.int32)
i2 = in_tensors[1].astype(np.int32)
out_tensor = qrec.scale_mul_biases_q.apply_scales(
op(i1, i2, np.int32))
else:
# larger scale should be scaled
qrec.set_add_scale()
if qrec.scaled_idx:
i1 = in_tensors[0].astype(np.int32)
i2 = qrec.scale_in_mul_biases_q.apply_scales(in_tensors[1])
else:
i1 = qrec.scale_in_mul_biases_q.apply_scales(in_tensors[0])
i2 = in_tensors[1].astype(np.int32)
out_tensor = qrec.scale_mul_biases_q.apply_scales(op(i1, i2, None))
if params.transpose_out:
out_tensor = np.transpose(out_tensor, params.transpose_out[0])
return qrec.get_outputs(params, [qrec.out_qs[0].clip(out_tensor)],
ktype="symmetric")
def execute(cls, params, in_tensors, qrec: QuantizationRecordBase,
**kwargs):
in_tensors = qrec.prepare_inputs(params, in_tensors, ktype="symmetric")
func = PIECEWISE_OPS[params.__class__]
op = func['op']
if func['is_mult']:
qrec.set_scale(in_idx=(0, 1), out_idx=0)
i1 = in_tensors[0].astype(np.int32)
i2 = in_tensors[1].astype(np.int32)
res = qrec.scale_mul_biases_q.apply_scales(op(i1, i2, np.int32))
else:
# larger scale should be scaled
qrec.set_add_scale()
if qrec.scaled_idx:
i1 = in_tensors[0].astype(np.int32)
i2 = qrec.scale_in_mul_biases_q.apply_scales(in_tensors[1])
else:
i1 = qrec.scale_in_mul_biases_q.apply_scales(in_tensors[0])
i2 = in_tensors[1].astype(np.int32)
res = qrec.scale_mul_biases_q.apply_scales(op(i1, i2, None))
return qrec.get_outputs(params, [qrec.out_qs[0].clip(res)],
ktype="symmetric")
