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