Open arnozxn opened 5 months ago
I replaced the backbone network with LSKNet, but the test result is abnormal in the log during training, and the map is even greater than 1. Does anyone know why this is? {"train_lr": 9.999999999999247e-05, "train_class_error": 72.48062624495238, "train_loss": 22.32403853104765, "train_loss_ce": 1.7056896129660755, "train_loss_bbox": 0.6676112288199127, "train_loss_giou": 1.3335254561607217, "train_loss_ce_0": 1.703107129561414, "train_loss_bbox_0": 0.6271042094230974, "train_loss_giou_0": 1.2954364228845823, "train_loss_ce_1": 1.7076059571847266, "train_loss_bbox_1": 0.6804102641614062, "train_loss_giou_1": 1.3498873070590125, "train_loss_ce_2": 1.7106536044243708, "train_loss_bbox_2": 0.6826548581706428, "train_loss_giou_2": 1.3508466516100328, "train_loss_ce_3": 1.7141013477276839, "train_loss_bbox_3": 0.691834336042247, "train_loss_giou_3": 1.3587410876318717, "train_loss_ce_4": 1.7086172861328346, "train_loss_bbox_4": 0.6833480554040678, "train_loss_giou_4": 1.3528637243979473, "train_loss_ce_unscaled": 0.8528448064830377, "train_class_error_unscaled": 72.48062624495238, "train_loss_bbox_unscaled": 0.13352224572979274, "train_loss_giou_unscaled": 0.6667627280803609, "train_cardinality_error_unscaled": 293.41843497962566, "train_loss_ce_0_unscaled": 0.851553564780707, "train_loss_bbox_0_unscaled": 0.12542084182615418, "train_loss_giou_0_unscaled": 0.6477182114422911, "train_cardinality_error_0_unscaled": 293.41843497962566, "train_loss_ce_1_unscaled": 0.8538029785923633, "train_loss_bbox_1_unscaled": 0.1360820527831822, "train_loss_giou_1_unscaled": 0.6749436535295062, "train_cardinality_error_1_unscaled": 293.41843497962566, "train_loss_ce_2_unscaled": 0.8553268022121854, "train_loss_bbox_2_unscaled": 0.13653097162379854, "train_loss_giou_2_unscaled": 0.6754233258050164, "train_cardinality_error_2_unscaled": 293.41843497962566, "train_loss_ce_3_unscaled": 0.8570506738638419, "train_loss_bbox_3_unscaled": 0.1383668671476913, "train_loss_giou_3_unscaled": 0.6793705438159359, "train_cardinality_error_3_unscaled": 293.41843497962566, "train_loss_ce_4_unscaled": 0.8543086430664173, "train_loss_bbox_4_unscaled": 0.13666961106358008, "train_loss_giou_4_unscaled": 0.6764318621989737, "train_cardinality_error_4_unscaled": 293.41843497962566, "test_class_error": 98.31917419586182, "test_loss": 36.67297601013183, "test_loss_ce": 2.2453352907657624, "test_loss_bbox": 4.80726374437809, "test_loss_giou": 1.744889467406273, "test_loss_ce_0": 1.7523636434555054, "test_loss_bbox_0": 1.7386644288897515, "test_loss_giou_0": 1.9242841378688813, "test_loss_ce_1": 1.8169942341327667, "test_loss_bbox_1": 2.5693576932430267, "test_loss_giou_1": 1.6476937449932099, "test_loss_ce_2": 1.7992777534484863, "test_loss_bbox_2": 2.9500436752319334, "test_loss_giou_2": 1.6473702851176262, "test_loss_ce_3": 1.7374481472969054, "test_loss_bbox_3": 1.5866199144124984, "test_loss_giou_3": 1.7275373920202255, "test_loss_ce_4": 1.7414990911006927, "test_loss_bbox_4": 1.5899762920856475, "test_loss_giou_4": 1.6463570453882217, "test_loss_ce_unscaled": 1.1226676453828812, "test_class_error_unscaled": 98.31917419586182, "test_loss_bbox_unscaled": 0.9614527473986149, "test_loss_giou_unscaled": 0.8724447337031365, "test_cardinality_error_unscaled": 292.733, "test_loss_ce_0_unscaled": 0.8761818217277527, "test_loss_bbox_0_unscaled": 0.3477328853532672, "test_loss_giou_0_unscaled": 0.9621420689344407, "test_cardinality_error_0_unscaled": 292.733, "test_loss_ce_1_unscaled": 0.9084971170663834, "test_loss_bbox_1_unscaled": 0.5138715385675431, "test_loss_giou_1_unscaled": 0.8238468724966049, "test_cardinality_error_1_unscaled": 292.733, "test_loss_ce_2_unscaled": 0.8996388767242431, "test_loss_bbox_2_unscaled": 0.590008734536171, "test_loss_giou_2_unscaled": 0.8236851425588131, "test_cardinality_error_2_unscaled": 292.733, "test_loss_ce_3_unscaled": 0.8687240736484527, "test_loss_bbox_3_unscaled": 0.31732398263961076, "test_loss_giou_3_unscaled": 0.8637686960101127, "test_cardinality_error_3_unscaled": 292.733, "test_loss_ce_4_unscaled": 0.8707495455503463, "test_loss_bbox_4_unscaled": 0.31799525859057903, "test_loss_giou_4_unscaled": 0.8231785226941108, "test_cardinality_error_4_unscaled": 292.733, "test_coco_eval_bbox": [1.8120953964383796e-07, 1.0375569190032606e-06, 4.669005873194366e-09, 0.0, 0.0, 2.9677487952357753e-07, 2.158273381294964e-05, 0.00016366906474820142, 0.0006241007194244604, 0.0, 0.0, 0.0010631127450980392], "epoch": 10, "n_parameters": 22865341}
I replaced the backbone network with LSKNet, but the test result is abnormal in the log during training, and the map is even greater than 1. Does anyone know why this is? {"train_lr": 9.999999999999247e-05, "train_class_error": 72.48062624495238, "train_loss": 22.32403853104765, "train_loss_ce": 1.7056896129660755, "train_loss_bbox": 0.6676112288199127, "train_loss_giou": 1.3335254561607217, "train_loss_ce_0": 1.703107129561414, "train_loss_bbox_0": 0.6271042094230974, "train_loss_giou_0": 1.2954364228845823, "train_loss_ce_1": 1.7076059571847266, "train_loss_bbox_1": 0.6804102641614062, "train_loss_giou_1": 1.3498873070590125, "train_loss_ce_2": 1.7106536044243708, "train_loss_bbox_2": 0.6826548581706428, "train_loss_giou_2": 1.3508466516100328, "train_loss_ce_3": 1.7141013477276839, "train_loss_bbox_3": 0.691834336042247, "train_loss_giou_3": 1.3587410876318717, "train_loss_ce_4": 1.7086172861328346, "train_loss_bbox_4": 0.6833480554040678, "train_loss_giou_4": 1.3528637243979473, "train_loss_ce_unscaled": 0.8528448064830377, "train_class_error_unscaled": 72.48062624495238, "train_loss_bbox_unscaled": 0.13352224572979274, "train_loss_giou_unscaled": 0.6667627280803609, "train_cardinality_error_unscaled": 293.41843497962566, "train_loss_ce_0_unscaled": 0.851553564780707, "train_loss_bbox_0_unscaled": 0.12542084182615418, "train_loss_giou_0_unscaled": 0.6477182114422911, "train_cardinality_error_0_unscaled": 293.41843497962566, "train_loss_ce_1_unscaled": 0.8538029785923633, "train_loss_bbox_1_unscaled": 0.1360820527831822, "train_loss_giou_1_unscaled": 0.6749436535295062, "train_cardinality_error_1_unscaled": 293.41843497962566, "train_loss_ce_2_unscaled": 0.8553268022121854, "train_loss_bbox_2_unscaled": 0.13653097162379854, "train_loss_giou_2_unscaled": 0.6754233258050164, "train_cardinality_error_2_unscaled": 293.41843497962566, "train_loss_ce_3_unscaled": 0.8570506738638419, "train_loss_bbox_3_unscaled": 0.1383668671476913, "train_loss_giou_3_unscaled": 0.6793705438159359, "train_cardinality_error_3_unscaled": 293.41843497962566, "train_loss_ce_4_unscaled": 0.8543086430664173, "train_loss_bbox_4_unscaled": 0.13666961106358008, "train_loss_giou_4_unscaled": 0.6764318621989737, "train_cardinality_error_4_unscaled": 293.41843497962566, "test_class_error": 98.31917419586182, "test_loss": 36.67297601013183, "test_loss_ce": 2.2453352907657624, "test_loss_bbox": 4.80726374437809, "test_loss_giou": 1.744889467406273, "test_loss_ce_0": 1.7523636434555054, "test_loss_bbox_0": 1.7386644288897515, "test_loss_giou_0": 1.9242841378688813, "test_loss_ce_1": 1.8169942341327667, "test_loss_bbox_1": 2.5693576932430267, "test_loss_giou_1": 1.6476937449932099, "test_loss_ce_2": 1.7992777534484863, "test_loss_bbox_2": 2.9500436752319334, "test_loss_giou_2": 1.6473702851176262, "test_loss_ce_3": 1.7374481472969054, "test_loss_bbox_3": 1.5866199144124984, "test_loss_giou_3": 1.7275373920202255, "test_loss_ce_4": 1.7414990911006927, "test_loss_bbox_4": 1.5899762920856475, "test_loss_giou_4": 1.6463570453882217, "test_loss_ce_unscaled": 1.1226676453828812, "test_class_error_unscaled": 98.31917419586182, "test_loss_bbox_unscaled": 0.9614527473986149, "test_loss_giou_unscaled": 0.8724447337031365, "test_cardinality_error_unscaled": 292.733, "test_loss_ce_0_unscaled": 0.8761818217277527, "test_loss_bbox_0_unscaled": 0.3477328853532672, "test_loss_giou_0_unscaled": 0.9621420689344407, "test_cardinality_error_0_unscaled": 292.733, "test_loss_ce_1_unscaled": 0.9084971170663834, "test_loss_bbox_1_unscaled": 0.5138715385675431, "test_loss_giou_1_unscaled": 0.8238468724966049, "test_cardinality_error_1_unscaled": 292.733, "test_loss_ce_2_unscaled": 0.8996388767242431, "test_loss_bbox_2_unscaled": 0.590008734536171, "test_loss_giou_2_unscaled": 0.8236851425588131, "test_cardinality_error_2_unscaled": 292.733, "test_loss_ce_3_unscaled": 0.8687240736484527, "test_loss_bbox_3_unscaled": 0.31732398263961076, "test_loss_giou_3_unscaled": 0.8637686960101127, "test_cardinality_error_3_unscaled": 292.733, "test_loss_ce_4_unscaled": 0.8707495455503463, "test_loss_bbox_4_unscaled": 0.31799525859057903, "test_loss_giou_4_unscaled": 0.8231785226941108, "test_cardinality_error_4_unscaled": 292.733, "test_coco_eval_bbox": [1.8120953964383796e-07, 1.0375569190032606e-06, 4.669005873194366e-09, 0.0, 0.0, 2.9677487952357753e-07, 2.158273381294964e-05, 0.00016366906474820142, 0.0006241007194244604, 0.0, 0.0, 0.0010631127450980392], "epoch": 10, "n_parameters": 22865341}