2025-07-01 06:48:22,133 INFO log will be stored in /home/zhutao/horizon_j6/model/hb_compile.log 2025-07-01 06:48:22,134 INFO Start hb_compile... 2025-07-01 06:48:22,134 INFO hbdk version: 4.1.17 2025-07-01 06:48:22,134 INFO hmct version: 2.1.9 2025-07-01 06:48:22,134 INFO hb_compile version: 3.3.11 2025-07-01 06:48:22,134 INFO Start verifying yaml 2025-07-01 06:48:22,391 INFO Using onnx model file: /home/zhutao/horizon_j6/model/bev_fushion/onnx/nv12/onnx_v3_1_nv12.onnx 2025-07-01 06:48:22,392 INFO Model has 3 inputs according to model file 2025-07-01 06:48:22,392 INFO Input shape [6, 270336] has length: 2, make sure it is a featuremap input 2025-07-01 06:48:22,392 INFO Input shape [40000, 4] has length: 2, make sure it is a featuremap input 2025-07-01 06:48:22,392 WARNING Input shape is not four-dimensional, input_layout_train should be featuremap 2025-07-01 06:48:22,392 WARNING Input shape is not four-dimensional, input_layout_train should be featuremap 2025-07-01 06:48:22,392 WARNING norm_type parameter is deprecated and will be determined by the configuration of mean/scale/std parameter 2025-07-01 06:48:22,392 WARNING Input node img_nv12's input_source not set, it will be set to None by default 2025-07-01 06:48:22,392 WARNING Input node feature's input_source not set, it will be set to None by default 2025-07-01 06:48:22,392 WARNING Input node onnx::Gather_3's input_source not set, it will be set to None by default 2025-07-01 06:48:22,392 INFO Parameter compile_mode is set to latency, balance_factor will set to 100. 2025-07-01 06:48:22,392 INFO End verifying yaml 2025-07-01 06:48:22,393 INFO Processing calibration set data for input[img_nv12] 2025-07-01 06:48:22,393 INFO ******************************************* 2025-07-01 06:48:22,393 INFO First calibration picture name: /home/zhutao/horizon_j6/model/data/export/nv12/img/frame_000000_img_nv12.npy 2025-07-01 06:48:22,405 INFO First calibration picture md5: e15ff4cf1ac130419763a8f394f8b680 2025-07-01 06:48:22,406 INFO ******************************************* 2025-07-01 06:48:22,463 INFO Finished. data num: 10 2025-07-01 06:48:22,463 INFO Processing calibration set data for input[feature] 2025-07-01 06:48:22,463 INFO ******************************************* 2025-07-01 06:48:22,463 INFO First calibration picture name: /home/zhutao/horizon_j6/model/data/export/nv12/feature/frame_000000_feature.npy 2025-07-01 06:48:22,522 INFO First calibration picture md5: 82852e7c6039bf7d1360375813c4b501 2025-07-01 06:48:22,522 INFO ******************************************* 2025-07-01 06:48:22,776 INFO Finished. data num: 10 2025-07-01 06:48:22,776 INFO Processing calibration set data for input[onnx::Gather_3] 2025-07-01 06:48:22,776 INFO ******************************************* 2025-07-01 06:48:22,776 INFO First calibration picture name: /home/zhutao/horizon_j6/model/data/export/nv12/coors/frame_000000_gather.npy 2025-07-01 06:48:22,778 INFO First calibration picture md5: 779a0f2a6a8e31be78dd02d64fddca9a 2025-07-01 06:48:22,778 INFO ******************************************* 2025-07-01 06:48:22,872 INFO Finished. data num: 10 2025-07-01 06:48:22,941 INFO Start to Horizon NN Model Convert. 2025-07-01 06:48:23,107 INFO Loading horizon_nn debug methods: {'dump_calibration_data'}. 2025-07-01 06:48:23,107 INFO The specified model compilation architecture: nash. 2025-07-01 06:48:23,107 INFO The specified model compilation optimization parameters: []. 2025-07-01 06:48:23,117 INFO Start to prepare the onnx model. 2025-07-01 06:48:23,148 INFO Input ONNX Model Information: ONNX IR version: 8 Opset version: ['ai.onnx v16', 'horizon v1'] Producer: pytorch v1.12.1 Domain: None Version: None Graph input: img_nv12: shape=[6, 270336], dtype=FLOAT32 feature: shape=[1, 10, 20, 40000], dtype=FLOAT32 ...Gather_3: shape=[40000, 4], dtype=INT32 Graph output: reg_0: shape=[1, 2, 120, 180], dtype=FLOAT32 height_0: shape=[1, 1, 120, 180], dtype=FLOAT32 dim_0: shape=[1, 3, 120, 180], dtype=FLOAT32 rot_0: shape=[1, 2, 120, 180], dtype=FLOAT32 vel_0: shape=[1, 2, 120, 180], dtype=FLOAT32 hm_0: shape=[1, 1, 120, 180], dtype=FLOAT32 reg_1: shape=[1, 2, 120, 180], dtype=FLOAT32 height_1: shape=[1, 1, 120, 180], dtype=FLOAT32 dim_1: shape=[1, 3, 120, 180], dtype=FLOAT32 rot_1: shape=[1, 2, 120, 180], dtype=FLOAT32 vel_1: shape=[1, 2, 120, 180], dtype=FLOAT32 hm_1: shape=[1, 1, 120, 180], dtype=FLOAT32 reg_2: shape=[1, 2, 120, 180], dtype=FLOAT32 height_2: shape=[1, 1, 120, 180], dtype=FLOAT32 dim_2: shape=[1, 3, 120, 180], dtype=FLOAT32 rot_2: shape=[1, 2, 120, 180], dtype=FLOAT32 vel_2: shape=[1, 2, 120, 180], dtype=FLOAT32 hm_2: shape=[1, 1, 120, 180], dtype=FLOAT32 reg_3: shape=[1, 2, 120, 180], dtype=FLOAT32 height_3: shape=[1, 1, 120, 180], dtype=FLOAT32 dim_3: shape=[1, 3, 120, 180], dtype=FLOAT32 rot_3: shape=[1, 2, 120, 180], dtype=FLOAT32 vel_3: shape=[1, 2, 120, 180], dtype=FLOAT32 hm_3: shape=[1, 1, 120, 180], dtype=FLOAT32 reg_4: shape=[1, 2, 120, 180], dtype=FLOAT32 height_4: shape=[1, 1, 120, 180], dtype=FLOAT32 dim_4: shape=[1, 3, 120, 180], dtype=FLOAT32 rot_4: shape=[1, 2, 120, 180], dtype=FLOAT32 vel_4: shape=[1, 2, 120, 180], dtype=FLOAT32 hm_4: shape=[1, 1, 120, 180], dtype=FLOAT32 2025-07-01 06:48:23,779 INFO The original model's opset version is 16, try converting to opset 19. 2025-07-01 06:48:24,428 INFO End to prepare the onnx model. 2025-07-01 06:48:24,715 INFO Saving model to: /home/zhutao/horizon_j6/model/bev_fushion/hbm/onnx_v3_1_nv12/onnx_v3_1_nv12_original_float_model.onnx. 2025-07-01 06:48:24,715 INFO Start to optimize the onnx model. 2025-07-01 06:48:28,063 INFO End to optimize the onnx model. 2025-07-01 06:48:28,363 INFO Saving model to: /home/zhutao/horizon_j6/model/bev_fushion/hbm/onnx_v3_1_nv12/onnx_v3_1_nv12_optimized_float_model.onnx. 2025-07-01 06:48:28,364 INFO The weight calibration parameters: bias_correction: {'num_sample': 1, 'metric': 'mse'} The layerwise search parameters: metric: mse All nodes in the model are set to datatype: int8 2025-07-01 06:48:28,364 INFO Start to calibrate the model. Warning: The input0 of Node(name:ScatterND_76, type:ScatterND) does not support data type: int8 Warning: The input2 of Node(name:ScatterND_76, type:ScatterND) does not support data type: int8 Warning: The input1 of Node(name:GridSample_308, type:GridSample) does not support data type: int8 Warning: The input1 of Node(name:GridSample_312, type:GridSample) does not support data type: int8 Warning: The input1 of Node(name:GridSample_316, type:GridSample) does not support data type: int8 Warning: The input1 of Node(name:GridSample_320, type:GridSample) does not support data type: int8 Warning: The input1 of Node(name:GridSample_324, type:GridSample) does not support data type: int8 Warning: The input1 of Node(name:GridSample_328, type:GridSample) does not support data type: int8 2025-07-01 06:48:28,979 INFO There are 10 samples in the dataset. 2025-07-01 06:48:29,417 INFO Provided calibration data md5: 99fa22c99b845d1cc5f12ca00bdce7ca 2025-07-01 06:48:29,417 INFO Run calibration model with layerwise search method. 2025-07-01 06:48:29,870 INFO Calibration using batch 1 calibration in progress: 100%|██████████████████████████████████████████████████████████████████████| 10/10 [01:39<00:00, 9.98s/it] 2025-07-01 06:50:29,234 INFO Calculate all sensitivity nodes with max calibration method using mse metric. 2025-07-01 06:50:29,332 INFO Start calculating activation sensitivity... 2025-07-01 06:50:30,586 INFO There are 1 samples in the dataset. Progress: 0%| | 0/143 [00:00