{ "per_channel": false, "reduce_range": false, "per_model_config": { "encoder_model": { "op_types": [ "Gather", "ReduceMean", "Add", "Pow", "Concat", "Erf", "Softmax", "Div", "Conv", "Transpose", "Mul", "Shape", "Sqrt", "MatMul", "Unsqueeze", "Reshape", "Sub", "Constant" ], "weight_type": "QUInt8" }, "decoder_model": { "op_types": [ "Concat", "Div", "Cast", "Slice", "Unsqueeze", "MatMul", "Constant", "Where", "Expand", "Softmax", "Shape", "Gather", "Pow", "ConstantOfShape", "Add", "Squeeze", "Equal", "Range", "Less", "Mul", "Sub", "ReduceMean", "Erf", "Transpose", "Sqrt", "Reshape" ], "weight_type": "QInt8" }, "decoder_with_past_model": { "op_types": [ "Gather", "ReduceMean", "Add", "Pow", "Concat", "Softmax", "Erf", "MatMul", "Div", "Slice", "Mul", "Unsqueeze", "Shape", "Sqrt", "Transpose", "Reshape", "Sub", "Constant" ], "weight_type": "QInt8" }, "decoder_model_merged": { "op_types": [ "Concat", "Div", "Cast", "Slice", "Unsqueeze", "MatMul", "Constant", "Where", "Expand", "Softmax", "Shape", "Gather", "Pow", "ConstantOfShape", "Add", "If", "Squeeze", "Equal", "Range", "Less", "Mul", "Sub", "ReduceMean", "Erf", "Transpose", "Sqrt", "Reshape" ], "weight_type": "QInt8" } } }