sima_qat.onnx_ops

Functions

fake_quantize_per_tensor_affine(g, inputs, scale, ...)
fake_dequantize_per_tensor_affine(g, inputs, scale, ...)
fake_quantize_per_channel_affine(g, inputs, scales, ...)

Module Contents

sima_qat.onnx_ops.fake_quantize_per_tensor_affine(g: torch.onnx._internal.jit_utils.GraphContext, inputs, scale, zero_point, quant_min=-128, quant_max=127, dtype=torch.dtype)

sima_qat.onnx_ops.fake_dequantize_per_tensor_affine(g: torch.onnx._internal.jit_utils.GraphContext, inputs, scale, zero_point, quant_min=-128, quant_max=127, dtype=torch.dtype)

sima_qat.onnx_ops.fake_quantize_per_channel_affine(g: torch.onnx._internal.jit_utils.GraphContext, inputs, scales, zero_points, axis, quant_min=-128, quant_max=127, dtype=torch.dtype)