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