Forums - Linux HTP run quantized mod

I run mobilenetv3 with CPU successfully. But I want to run quantized model falied, I got error: 0.0ms [ INFO ] QnnDsp QnnTensor_createGraphTensor started, graph = 0x1 0.0ms [ INFO ] QnnDsp QnnTensor_createGraphTensor graph 1 0.0ms [ INFO ] QnnDsp QnnTensor_createGraphTensor done. graph = 0x1, status 0x0 0.0ms [ INFO ] QnnDsp QnnBackend_validateOpConfig started, backend = 0x1 0.0ms [ ERROR ] QnnDsp Input[0] has incorrect Value 1032, expected < 1032. 0.0ms [ ERROR ] QnnDsp validateNativeOps _blocks_blocks_3_blocks_3_1_se_Mul:qti.aisw:ElementWiseMultiply op validator (quantized and FP16) failed 3110 and 3110 0.0ms [ ERROR ] QnnDsp QnnBackend_validateOpConfig failed 3110 0.0ms [ ERROR ] QnnDsp Failed to validate op _blocks_blocks_3_blocks_3_1_se_Mul with error 0xc26 [ ERROR ] QnnModel::addNode() validating node _blocks_blocks_3_blocks_3_1_se_Mul failed. [ ERROR ] model.addNode(QNN_OPCONFIG_VERSION_1, "_blocks_blocks_3_blocks_3_1_se_Mul", "qti.aisw", "ElementWiseMultiply", nullptr, 0, inputs__blocks_blocks_3_blocks_3_1_se_Mul, 2, outputs__blocks_blocks_3_blocks_3_1_se_Mul, 1 ) expected MODEL_NO_ERROR, got MODEL_GRAPH_ERROR [ ERROR ] addNode__blocks_blocks_3_blocks_3_1_se_Mul(mobilenetv3_small_050_Opset16_quant) expected MODEL_NO_ERROR, got MODEL_GRAPH_ERROR 175.0ms [ ERROR ] Failed in composeGraphs() 175.0ms [ ERROR ] ComposeGraphs Failed with error = 1 Graph Prepare failure The input[0] in *_quant_net.json: "x": { "id": 1, "type": 0, "dataFormat": 0, "data_type": 1032, "unquantized_data_type": 562, "axis_format": "NHWC", "src_axis_format": "NCHW", "quant_params": { "definition": 1, "encoding": 0, "is_overridden": false, "scale_offset": { "bitwidth": 8, "minimum": 0.0, "maximum": 255.0, "scale": 1.0, "offset": 0, "is_symmetric": false, "is_fixed_point": true } }, "dims": [ 1, 224, 224, 3 ], "is_quantizable": true } }, The input I have save int8 binary data, does it not right?