ppocr detv3检测模型转换出现问题

[zhangshabao]

检测模型转换时候出现
Traceback (most recent call last):
File "f:/2023_work/OCR/gitclone_code/PaddleOCR2Pytorch-main/converter/ch_ppocr_v3_det_converter.py", line 52, in load_paddle_weights
KeyError: 'Student2.backbone.conv.weight'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
converter = PPOCRv3DetConverter(cfg, paddle_pretrained_model_path)
File "f:/2023_work/OCR/gitclone_code/PaddleOCR2Pytorch-main/converter/ch_ppocr_v3_det_converter.py", line 13, in init
self.load_paddle_weights(paddle_pretrained_model_path)
File "f:/2023_work/OCR/gitclone_code/PaddleOCR2Pytorch-main/converter/ch_ppocr_v3_det_converter.py", line 55, in load_paddle_weights
print('paddle: {}, {}'.format(ppname, para_state_dict[ppname].shape))
KeyError: 'Student2.backbone.conv.weight'
这个要怎么解决

[frotms]

@zhangshabao
转换命令是？

[zhangshabao]

我是用ch_ppocr_v3_det_converter.py 这个直接python ch_ppocr_v3_det_converter.py

[frotms]

python ./converter/ch_ppocr_v3_det_converter.py --src_model_path paddle_ch_PP-OCRv3_rec_train_dir

[zhangshabao]

paddle: backbone.conv.conv.weight ---- (8, 3, 3, 3)
paddle: backbone.conv.bn.weight ---- (8,)
paddle: backbone.conv.bn.bias ---- (8,)
paddle: backbone.conv.bn._mean ---- (8,)
paddle: backbone.conv.bn._variance ---- (8,)
paddle: backbone.stage0.0.expand_conv.conv.weight ---- (8, 8, 1, 1)
paddle: backbone.stage0.0.expand_conv.bn.weight ---- (8,)
paddle: backbone.stage0.0.expand_conv.bn.bias ---- (8,)
paddle: backbone.stage0.0.expand_conv.bn._mean ---- (8,)
paddle: backbone.stage0.0.expand_conv.bn._variance ---- (8,)
paddle: backbone.stage0.0.bottleneck_conv.conv.weight ---- (8, 1, 3, 3)
paddle: backbone.stage0.0.bottleneck_conv.bn.weight ---- (8,)
paddle: backbone.stage0.0.bottleneck_conv.bn.bias ---- (8,)
paddle: backbone.stage0.0.bottleneck_conv.bn._mean ---- (8,)
paddle: backbone.stage0.0.bottleneck_conv.bn._variance ---- (8,)
paddle: backbone.stage0.0.linear_conv.conv.weight ---- (8, 8, 1, 1)
paddle: backbone.stage0.0.linear_conv.bn.weight ---- (8,)
paddle: backbone.stage0.0.linear_conv.bn.bias ---- (8,)
paddle: backbone.stage0.0.linear_conv.bn._mean ---- (8,)
paddle: backbone.stage0.0.linear_conv.bn._variance ---- (8,)
paddle: backbone.stage1.0.expand_conv.conv.weight ---- (40, 8, 1, 1)
paddle: backbone.stage1.0.expand_conv.bn.weight ---- (40,)
paddle: backbone.stage1.0.expand_conv.bn.bias ---- (40,)
paddle: backbone.stage1.0.expand_conv.bn._mean ---- (40,)
paddle: backbone.stage1.0.expand_conv.bn._variance ---- (40,)
paddle: backbone.stage1.0.bottleneck_conv.conv.weight ---- (40, 1, 3, 3)
paddle: backbone.stage1.0.bottleneck_conv.bn.weight ---- (40,)
paddle: backbone.stage1.0.bottleneck_conv.bn.bias ---- (40,)
paddle: backbone.stage1.0.bottleneck_conv.bn._mean ---- (40,)
paddle: backbone.stage1.0.bottleneck_conv.bn._variance ---- (40,)
paddle: backbone.stage1.0.linear_conv.conv.weight ---- (16, 40, 1, 1)
paddle: backbone.stage1.0.linear_conv.bn.weight ---- (16,)
paddle: backbone.stage1.0.linear_conv.bn.bias ---- (16,)
paddle: backbone.stage1.0.linear_conv.bn._mean ---- (16,)
paddle: backbone.stage1.0.linear_conv.bn._variance ---- (16,)
paddle: backbone.stage1.1.expand_conv.conv.weight ---- (48, 16, 1, 1)
paddle: backbone.stage1.1.expand_conv.bn.weight ---- (48,)
paddle: backbone.stage1.1.expand_conv.bn.bias ---- (48,)
paddle: backbone.stage1.1.expand_conv.bn._mean ---- (48,)
paddle: backbone.stage1.1.expand_conv.bn._variance ---- (48,)
paddle: backbone.stage1.1.bottleneck_conv.conv.weight ---- (48, 1, 3, 3)
paddle: backbone.stage1.1.bottleneck_conv.bn.weight ---- (48,)
paddle: backbone.stage1.1.bottleneck_conv.bn.bias ---- (48,)
paddle: backbone.stage1.1.bottleneck_conv.bn._mean ---- (48,)
paddle: backbone.stage1.1.bottleneck_conv.bn._variance ---- (48,)
paddle: backbone.stage1.1.linear_conv.conv.weight ---- (16, 48, 1, 1)
paddle: backbone.stage1.1.linear_conv.bn.weight ---- (16,)
paddle: backbone.stage1.1.linear_conv.bn.bias ---- (16,)
paddle: backbone.stage1.1.linear_conv.bn._mean ---- (16,)
paddle: backbone.stage1.1.linear_conv.bn._variance ---- (16,)
paddle: backbone.stage2.0.expand_conv.conv.weight ---- (48, 16, 1, 1)
paddle: backbone.stage2.0.expand_conv.bn.weight ---- (48,)
paddle: backbone.stage2.0.expand_conv.bn.bias ---- (48,)
paddle: backbone.stage2.0.expand_conv.bn._mean ---- (48,)
paddle: backbone.stage2.0.expand_conv.bn._variance ---- (48,)
paddle: backbone.stage2.0.bottleneck_conv.conv.weight ---- (48, 1, 5, 5)
paddle: backbone.stage2.0.bottleneck_conv.bn.weight ---- (48,)
paddle: backbone.stage2.0.bottleneck_conv.bn.bias ---- (48,)
paddle: backbone.stage2.0.bottleneck_conv.bn._mean ---- (48,)
paddle: backbone.stage2.0.bottleneck_conv.bn._variance ---- (48,)
paddle: backbone.stage2.0.linear_conv.conv.weight ---- (24, 48, 1, 1)
paddle: backbone.stage2.0.linear_conv.bn.weight ---- (24,)
paddle: backbone.stage2.0.linear_conv.bn.bias ---- (24,)
paddle: backbone.stage2.0.linear_conv.bn._mean ---- (24,)
paddle: backbone.stage2.0.linear_conv.bn._variance ---- (24,)
paddle: backbone.stage2.1.expand_conv.conv.weight ---- (120, 24, 1, 1)
paddle: backbone.stage2.1.expand_conv.bn.weight ---- (120,)
paddle: backbone.stage2.1.expand_conv.bn.bias ---- (120,)
paddle: backbone.stage2.1.expand_conv.bn._mean ---- (120,)
paddle: backbone.stage2.1.expand_conv.bn._variance ---- (120,)
paddle: backbone.stage2.1.bottleneck_conv.conv.weight ---- (120, 1, 5, 5)
paddle: backbone.stage2.1.bottleneck_conv.bn.weight ---- (120,)
paddle: backbone.stage2.1.bottleneck_conv.bn.bias ---- (120,)
paddle: backbone.stage2.1.bottleneck_conv.bn._mean ---- (120,)
paddle: backbone.stage2.1.bottleneck_conv.bn._variance ---- (120,)
paddle: backbone.stage2.1.linear_conv.conv.weight ---- (24, 120, 1, 1)
paddle: backbone.stage2.1.linear_conv.bn.weight ---- (24,)
paddle: backbone.stage2.1.linear_conv.bn.bias ---- (24,)
paddle: backbone.stage2.1.linear_conv.bn._mean ---- (24,)
paddle: backbone.stage2.1.linear_conv.bn._variance ---- (24,)
paddle: backbone.stage2.2.expand_conv.conv.weight ---- (120, 24, 1, 1)
paddle: backbone.stage2.2.expand_conv.bn.weight ---- (120,)
paddle: backbone.stage2.2.expand_conv.bn.bias ---- (120,)
paddle: backbone.stage2.2.expand_conv.bn._mean ---- (120,)
paddle: backbone.stage2.2.expand_conv.bn._variance ---- (120,)
paddle: backbone.stage2.2.bottleneck_conv.conv.weight ---- (120, 1, 5, 5)
paddle: backbone.stage2.2.bottleneck_conv.bn.weight ---- (120,)
paddle: backbone.stage2.2.bottleneck_conv.bn.bias ---- (120,)
paddle: backbone.stage2.2.bottleneck_conv.bn._mean ---- (120,)
paddle: backbone.stage2.2.bottleneck_conv.bn._variance ---- (120,)
paddle: backbone.stage2.2.linear_conv.conv.weight ---- (24, 120, 1, 1)
paddle: backbone.stage2.2.linear_conv.bn.weight ---- (24,)
paddle: backbone.stage2.2.linear_conv.bn.bias ---- (24,)
paddle: backbone.stage2.2.linear_conv.bn._mean ---- (24,)
paddle: backbone.stage2.2.linear_conv.bn._variance ---- (24,)
paddle: backbone.stage2.3.expand_conv.conv.weight ---- (64, 24, 1, 1)
paddle: backbone.stage2.3.expand_conv.bn.weight ---- (64,)
paddle: backbone.stage2.3.expand_conv.bn.bias ---- (64,)
paddle: backbone.stage2.3.expand_conv.bn._mean ---- (64,)
paddle: backbone.stage2.3.expand_conv.bn._variance ---- (64,)
paddle: backbone.stage2.3.bottleneck_conv.conv.weight ---- (64, 1, 5, 5)
paddle: backbone.stage2.3.bottleneck_conv.bn.weight ---- (64,)
paddle: backbone.stage2.3.bottleneck_conv.bn.bias ---- (64,)
paddle: backbone.stage2.3.bottleneck_conv.bn._mean ---- (64,)
paddle: backbone.stage2.3.bottleneck_conv.bn._variance ---- (64,)
paddle: backbone.stage2.3.linear_conv.conv.weight ---- (24, 64, 1, 1)
paddle: backbone.stage2.3.linear_conv.bn.weight ---- (24,)
paddle: backbone.stage2.3.linear_conv.bn.bias ---- (24,)
paddle: backbone.stage2.3.linear_conv.bn._mean ---- (24,)
paddle: backbone.stage2.3.linear_conv.bn._variance ---- (24,)
paddle: backbone.stage2.4.expand_conv.conv.weight ---- (72, 24, 1, 1)
paddle: backbone.stage2.4.expand_conv.bn.weight ---- (72,)
paddle: backbone.stage2.4.expand_conv.bn.bias ---- (72,)
paddle: backbone.stage2.4.expand_conv.bn._mean ---- (72,)
paddle: backbone.stage2.4.expand_conv.bn._variance ---- (72,)
paddle: backbone.stage2.4.bottleneck_conv.conv.weight ---- (72, 1, 5, 5)
paddle: backbone.stage2.4.bottleneck_conv.bn.weight ---- (72,)
paddle: backbone.stage2.4.bottleneck_conv.bn.bias ---- (72,)
paddle: backbone.stage2.4.bottleneck_conv.bn._mean ---- (72,)
paddle: backbone.stage2.4.bottleneck_conv.bn._variance ---- (72,)
paddle: backbone.stage2.4.linear_conv.conv.weight ---- (24, 72, 1, 1)
paddle: backbone.stage2.4.linear_conv.bn.weight ---- (24,)
paddle: backbone.stage2.4.linear_conv.bn.bias ---- (24,)
paddle: backbone.stage2.4.linear_conv.bn._mean ---- (24,)
paddle: backbone.stage2.4.linear_conv.bn._variance ---- (24,)
paddle: backbone.stage3.0.expand_conv.conv.weight ---- (144, 24, 1, 1)
paddle: backbone.stage3.0.expand_conv.bn.weight ---- (144,)
paddle: backbone.stage3.0.expand_conv.bn.bias ---- (144,)
paddle: backbone.stage3.0.expand_conv.bn._mean ---- (144,)
paddle: backbone.stage3.0.expand_conv.bn._variance ---- (144,)
paddle: backbone.stage3.0.bottleneck_conv.conv.weight ---- (144, 1, 5, 5)
paddle: backbone.stage3.0.bottleneck_conv.bn.weight ---- (144,)
paddle: backbone.stage3.0.bottleneck_conv.bn.bias ---- (144,)
paddle: backbone.stage3.0.bottleneck_conv.bn._mean ---- (144,)
paddle: backbone.stage3.0.bottleneck_conv.bn._variance ---- (144,)
paddle: backbone.stage3.0.linear_conv.conv.weight ---- (48, 144, 1, 1)
paddle: backbone.stage3.0.linear_conv.bn.weight ---- (48,)
paddle: backbone.stage3.0.linear_conv.bn.bias ---- (48,)
paddle: backbone.stage3.0.linear_conv.bn._mean ---- (48,)
paddle: backbone.stage3.0.linear_conv.bn._variance ---- (48,)
paddle: backbone.stage3.1.expand_conv.conv.weight ---- (288, 48, 1, 1)
paddle: backbone.stage3.1.expand_conv.bn.weight ---- (288,)
paddle: backbone.stage3.1.expand_conv.bn.bias ---- (288,)
paddle: backbone.stage3.1.expand_conv.bn._mean ---- (288,)
paddle: backbone.stage3.1.expand_conv.bn._variance ---- (288,)
paddle: backbone.stage3.1.bottleneck_conv.conv.weight ---- (288, 1, 5, 5)
paddle: backbone.stage3.1.bottleneck_conv.bn.weight ---- (288,)
paddle: backbone.stage3.1.bottleneck_conv.bn.bias ---- (288,)
paddle: backbone.stage3.1.bottleneck_conv.bn._mean ---- (288,)
paddle: backbone.stage3.1.bottleneck_conv.bn._variance ---- (288,)
paddle: backbone.stage3.1.linear_conv.conv.weight ---- (48, 288, 1, 1)
paddle: backbone.stage3.1.linear_conv.bn.weight ---- (48,)
paddle: backbone.stage3.1.linear_conv.bn.bias ---- (48,)
paddle: backbone.stage3.1.linear_conv.bn._mean ---- (48,)
paddle: backbone.stage3.1.linear_conv.bn._variance ---- (48,)
paddle: backbone.stage3.2.expand_conv.conv.weight ---- (288, 48, 1, 1)
paddle: backbone.stage3.2.expand_conv.bn.weight ---- (288,)
paddle: backbone.stage3.2.expand_conv.bn.bias ---- (288,)
paddle: backbone.stage3.2.expand_conv.bn._mean ---- (288,)
paddle: backbone.stage3.2.expand_conv.bn._variance ---- (288,)
paddle: backbone.stage3.2.bottleneck_conv.conv.weight ---- (288, 1, 5, 5)
paddle: backbone.stage3.2.bottleneck_conv.bn.weight ---- (288,)
paddle: backbone.stage3.2.bottleneck_conv.bn.bias ---- (288,)
paddle: backbone.stage3.2.bottleneck_conv.bn._mean ---- (288,)
paddle: backbone.stage3.2.bottleneck_conv.bn._variance ---- (288,)
paddle: backbone.stage3.2.linear_conv.conv.weight ---- (48, 288, 1, 1)
paddle: backbone.stage3.2.linear_conv.bn.weight ---- (48,)
paddle: backbone.stage3.2.linear_conv.bn.bias ---- (48,)
paddle: backbone.stage3.2.linear_conv.bn._mean ---- (48,)
paddle: backbone.stage3.2.linear_conv.bn._variance ---- (48,)
paddle: backbone.stage3.3.conv.weight ---- (288, 48, 1, 1)
paddle: backbone.stage3.3.bn.weight ---- (288,)
paddle: backbone.stage3.3.bn.bias ---- (288,)
paddle: backbone.stage3.3.bn._mean ---- (288,)
paddle: backbone.stage3.3.bn._variance ---- (288,)
paddle: neck.in2_conv.weight ---- (64, 8, 1, 1)
paddle: neck.in3_conv.weight ---- (64, 16, 1, 1)
paddle: neck.in4_conv.weight ---- (64, 24, 1, 1)
paddle: neck.in5_conv.weight ---- (64, 288, 1, 1)
paddle: neck.p5_conv.weight ---- (16, 64, 3, 3)
paddle: neck.p4_conv.weight ---- (16, 64, 3, 3)
paddle: neck.p3_conv.weight ---- (16, 64, 3, 3)
paddle: neck.p2_conv.weight ---- (16, 64, 3, 3)
paddle: neck.asf.conv.weight ---- (16, 64, 3, 3)
paddle: neck.asf.conv.bias ---- (16,)
paddle: neck.asf.spatial_scale.0.weight ---- (1, 1, 3, 3)
paddle: neck.asf.spatial_scale.2.weight ---- (1, 1, 1, 1)
paddle: neck.asf.channel_scale.0.weight ---- (4, 16, 1, 1)
paddle: head.binarize.conv1.weight ---- (16, 64, 3, 3)
paddle: head.binarize.conv_bn1.weight ---- (16,)
paddle: head.binarize.conv_bn1.bias ---- (16,)
paddle: head.binarize.conv_bn1._mean ---- (16,)
paddle: head.binarize.conv_bn1._variance ---- (16,)
paddle: head.binarize.conv2.weight ---- (16, 16, 2, 2)
paddle: head.binarize.conv2.bias ---- (16,)
paddle: head.binarize.conv_bn2.weight ---- (16,)
paddle: head.binarize.conv_bn2.bias ---- (16,)
paddle: head.binarize.conv_bn2._mean ---- (16,)
paddle: head.binarize.conv_bn2._variance ---- (16,)
paddle: head.binarize.conv3.weight ---- (16, 1, 2, 2)
paddle: head.binarize.conv3.bias ---- (1,)
paddle: head.thresh.conv1.weight ---- (16, 64, 3, 3)
paddle: head.thresh.conv_bn1.weight ---- (16,)
paddle: head.thresh.conv_bn1.bias ---- (16,)
paddle: head.thresh.conv_bn1._mean ---- (16,)
paddle: head.thresh.conv_bn1._variance ---- (16,)
paddle: head.thresh.conv2.weight ---- (16, 16, 2, 2)
paddle: head.thresh.conv2.bias ---- (16,)
paddle: head.thresh.conv_bn2.weight ---- (16,)
paddle: head.thresh.conv_bn2.bias ---- (16,)
paddle: head.thresh.conv_bn2._mean ---- (16,)
paddle: head.thresh.conv_bn2.variance ---- (16,)
paddle: head.thresh.conv3.weight ---- (16, 1, 2, 2)
paddle: head.thresh.conv3.bias ---- (1,)
pytorch: backbone.conv.conv.weight ---- torch.Size([8, 3, 3, 3])
pytorch: backbone.conv.bn.weight ---- torch.Size([8])
pytorch: backbone.conv.bn.bias ---- torch.Size([8])
pytorch: backbone.conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.0.expand_conv.conv.weight ---- torch.Size([8, 8, 1, 1])
pytorch: backbone.stages.0.0.expand_conv.bn.weight ---- torch.Size([8])
pytorch: backbone.stages.0.0.expand_conv.bn.bias ---- torch.Size([8])
pytorch: backbone.stages.0.0.expand_conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.stages.0.0.expand_conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.stages.0.0.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.0.bottleneck_conv.conv.weight ---- torch.Size([8, 1, 3, 3])
pytorch: backbone.stages.0.0.bottleneck_conv.bn.weight ---- torch.Size([8])
pytorch: backbone.stages.0.0.bottleneck_conv.bn.bias ---- torch.Size([8])
pytorch: backbone.stages.0.0.bottleneck_conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.stages.0.0.bottleneck_conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.stages.0.0.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.0.mid_se.conv1.weight ---- torch.Size([2, 8, 1, 1])
pytorch: backbone.stages.0.0.mid_se.conv1.bias ---- torch.Size([2])
pytorch: backbone.stages.0.0.mid_se.conv2.weight ---- torch.Size([8, 2, 1, 1])
pytorch: backbone.stages.0.0.mid_se.conv2.bias ---- torch.Size([8])
pytorch: backbone.stages.0.0.linear_conv.conv.weight ---- torch.Size([8, 8, 1, 1])
pytorch: backbone.stages.0.0.linear_conv.bn.weight ---- torch.Size([8])
pytorch: backbone.stages.0.0.linear_conv.bn.bias ---- torch.Size([8])
pytorch: backbone.stages.0.0.linear_conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.stages.0.0.linear_conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.stages.0.0.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.1.expand_conv.conv.weight ---- torch.Size([24, 8, 1, 1])
pytorch: backbone.stages.0.1.expand_conv.bn.weight ---- torch.Size([24])
pytorch: backbone.stages.0.1.expand_conv.bn.bias ---- torch.Size([24])
pytorch: backbone.stages.0.1.expand_conv.bn.running_mean ---- torch.Size([24])
pytorch: backbone.stages.0.1.expand_conv.bn.running_var ---- torch.Size([24])
pytorch: backbone.stages.0.1.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.1.bottleneck_conv.conv.weight ---- torch.Size([24, 1, 3, 3])
pytorch: backbone.stages.0.1.bottleneck_conv.bn.weight ---- torch.Size([24])
pytorch: backbone.stages.0.1.bottleneck_conv.bn.bias ---- torch.Size([24])
pytorch: backbone.stages.0.1.bottleneck_conv.bn.running_mean ---- torch.Size([24])
pytorch: backbone.stages.0.1.bottleneck_conv.bn.running_var ---- torch.Size([24])
pytorch: backbone.stages.0.1.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.1.linear_conv.conv.weight ---- torch.Size([8, 24, 1, 1])
pytorch: backbone.stages.0.1.linear_conv.bn.weight ---- torch.Size([8])
pytorch: backbone.stages.0.1.linear_conv.bn.bias ---- torch.Size([8])
pytorch: backbone.stages.0.1.linear_conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.stages.0.1.linear_conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.stages.0.1.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.2.expand_conv.conv.weight ---- torch.Size([32, 8, 1, 1])
pytorch: backbone.stages.0.2.expand_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.0.2.expand_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.0.2.expand_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.0.2.expand_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.0.2.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.2.bottleneck_conv.conv.weight ---- torch.Size([32, 1, 3, 3])
pytorch: backbone.stages.0.2.bottleneck_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.0.2.bottleneck_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.0.2.bottleneck_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.0.2.bottleneck_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.0.2.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.0.2.linear_conv.conv.weight ---- torch.Size([8, 32, 1, 1])
pytorch: backbone.stages.0.2.linear_conv.bn.weight ---- torch.Size([8])
pytorch: backbone.stages.0.2.linear_conv.bn.bias ---- torch.Size([8])
pytorch: backbone.stages.0.2.linear_conv.bn.running_mean ---- torch.Size([8])
pytorch: backbone.stages.0.2.linear_conv.bn.running_var ---- torch.Size([8])
pytorch: backbone.stages.0.2.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.0.expand_conv.conv.weight ---- torch.Size([32, 8, 1, 1])
pytorch: backbone.stages.1.0.expand_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.1.0.expand_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.1.0.expand_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.1.0.expand_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.1.0.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.0.bottleneck_conv.conv.weight ---- torch.Size([32, 1, 5, 5])
pytorch: backbone.stages.1.0.bottleneck_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.1.0.bottleneck_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.1.0.bottleneck_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.1.0.bottleneck_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.1.0.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.0.mid_se.conv1.weight ---- torch.Size([8, 32, 1, 1])
pytorch: backbone.stages.1.0.mid_se.conv1.bias ---- torch.Size([8])
pytorch: backbone.stages.1.0.mid_se.conv2.weight ---- torch.Size([32, 8, 1, 1])
pytorch: backbone.stages.1.0.mid_se.conv2.bias ---- torch.Size([32])
pytorch: backbone.stages.1.0.linear_conv.conv.weight ---- torch.Size([16, 32, 1, 1])
pytorch: backbone.stages.1.0.linear_conv.bn.weight ---- torch.Size([16])
pytorch: backbone.stages.1.0.linear_conv.bn.bias ---- torch.Size([16])
pytorch: backbone.stages.1.0.linear_conv.bn.running_mean ---- torch.Size([16])
pytorch: backbone.stages.1.0.linear_conv.bn.running_var ---- torch.Size([16])
pytorch: backbone.stages.1.0.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.1.expand_conv.conv.weight ---- torch.Size([88, 16, 1, 1])
pytorch: backbone.stages.1.1.expand_conv.bn.weight ---- torch.Size([88])
pytorch: backbone.stages.1.1.expand_conv.bn.bias ---- torch.Size([88])
pytorch: backbone.stages.1.1.expand_conv.bn.running_mean ---- torch.Size([88])
pytorch: backbone.stages.1.1.expand_conv.bn.running_var ---- torch.Size([88])
pytorch: backbone.stages.1.1.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.1.bottleneck_conv.conv.weight ---- torch.Size([88, 1, 5, 5])
pytorch: backbone.stages.1.1.bottleneck_conv.bn.weight ---- torch.Size([88])
pytorch: backbone.stages.1.1.bottleneck_conv.bn.bias ---- torch.Size([88])
pytorch: backbone.stages.1.1.bottleneck_conv.bn.running_mean ---- torch.Size([88])
pytorch: backbone.stages.1.1.bottleneck_conv.bn.running_var ---- torch.Size([88])
pytorch: backbone.stages.1.1.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.1.mid_se.conv1.weight ---- torch.Size([22, 88, 1, 1])
pytorch: backbone.stages.1.1.mid_se.conv1.bias ---- torch.Size([22])
pytorch: backbone.stages.1.1.mid_se.conv2.weight ---- torch.Size([88, 22, 1, 1])
pytorch: backbone.stages.1.1.mid_se.conv2.bias ---- torch.Size([88])
pytorch: backbone.stages.1.1.linear_conv.conv.weight ---- torch.Size([16, 88, 1, 1])
pytorch: backbone.stages.1.1.linear_conv.bn.weight ---- torch.Size([16])
pytorch: backbone.stages.1.1.linear_conv.bn.bias ---- torch.Size([16])
pytorch: backbone.stages.1.1.linear_conv.bn.running_mean ---- torch.Size([16])
pytorch: backbone.stages.1.1.linear_conv.bn.running_var ---- torch.Size([16])
pytorch: backbone.stages.1.1.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.2.expand_conv.conv.weight ---- torch.Size([88, 16, 1, 1])
pytorch: backbone.stages.1.2.expand_conv.bn.weight ---- torch.Size([88])
pytorch: backbone.stages.1.2.expand_conv.bn.bias ---- torch.Size([88])
pytorch: backbone.stages.1.2.expand_conv.bn.running_mean ---- torch.Size([88])
pytorch: backbone.stages.1.2.expand_conv.bn.running_var ---- torch.Size([88])
pytorch: backbone.stages.1.2.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.2.bottleneck_conv.conv.weight ---- torch.Size([88, 1, 5, 5])
pytorch: backbone.stages.1.2.bottleneck_conv.bn.weight ---- torch.Size([88])
pytorch: backbone.stages.1.2.bottleneck_conv.bn.bias ---- torch.Size([88])
pytorch: backbone.stages.1.2.bottleneck_conv.bn.running_mean ---- torch.Size([88])
pytorch: backbone.stages.1.2.bottleneck_conv.bn.running_var ---- torch.Size([88])
pytorch: backbone.stages.1.2.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.2.mid_se.conv1.weight ---- torch.Size([22, 88, 1, 1])
pytorch: backbone.stages.1.2.mid_se.conv1.bias ---- torch.Size([22])
pytorch: backbone.stages.1.2.mid_se.conv2.weight ---- torch.Size([88, 22, 1, 1])
pytorch: backbone.stages.1.2.mid_se.conv2.bias ---- torch.Size([88])
pytorch: backbone.stages.1.2.linear_conv.conv.weight ---- torch.Size([16, 88, 1, 1])
pytorch: backbone.stages.1.2.linear_conv.bn.weight ---- torch.Size([16])
pytorch: backbone.stages.1.2.linear_conv.bn.bias ---- torch.Size([16])
pytorch: backbone.stages.1.2.linear_conv.bn.running_mean ---- torch.Size([16])
pytorch: backbone.stages.1.2.linear_conv.bn.running_var ---- torch.Size([16])
pytorch: backbone.stages.1.2.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.3.expand_conv.conv.weight ---- torch.Size([40, 16, 1, 1])
pytorch: backbone.stages.1.3.expand_conv.bn.weight ---- torch.Size([40])
pytorch: backbone.stages.1.3.expand_conv.bn.bias ---- torch.Size([40])
pytorch: backbone.stages.1.3.expand_conv.bn.running_mean ---- torch.Size([40])
pytorch: backbone.stages.1.3.expand_conv.bn.running_var ---- torch.Size([40])
pytorch: backbone.stages.1.3.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.3.bottleneck_conv.conv.weight ---- torch.Size([40, 1, 5, 5])
pytorch: backbone.stages.1.3.bottleneck_conv.bn.weight ---- torch.Size([40])
pytorch: backbone.stages.1.3.bottleneck_conv.bn.bias ---- torch.Size([40])
pytorch: backbone.stages.1.3.bottleneck_conv.bn.running_mean ---- torch.Size([40])
pytorch: backbone.stages.1.3.bottleneck_conv.bn.running_var ---- torch.Size([40])
pytorch: backbone.stages.1.3.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.3.mid_se.conv1.weight ---- torch.Size([10, 40, 1, 1])
pytorch: backbone.stages.1.3.mid_se.conv1.bias ---- torch.Size([10])
pytorch: backbone.stages.1.3.mid_se.conv2.weight ---- torch.Size([40, 10, 1, 1])
pytorch: backbone.stages.1.3.mid_se.conv2.bias ---- torch.Size([40])
pytorch: backbone.stages.1.3.linear_conv.conv.weight ---- torch.Size([16, 40, 1, 1])
pytorch: backbone.stages.1.3.linear_conv.bn.weight ---- torch.Size([16])
pytorch: backbone.stages.1.3.linear_conv.bn.bias ---- torch.Size([16])
pytorch: backbone.stages.1.3.linear_conv.bn.running_mean ---- torch.Size([16])
pytorch: backbone.stages.1.3.linear_conv.bn.running_var ---- torch.Size([16])
pytorch: backbone.stages.1.3.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.4.expand_conv.conv.weight ---- torch.Size([48, 16, 1, 1])
pytorch: backbone.stages.1.4.expand_conv.bn.weight ---- torch.Size([48])
pytorch: backbone.stages.1.4.expand_conv.bn.bias ---- torch.Size([48])
pytorch: backbone.stages.1.4.expand_conv.bn.running_mean ---- torch.Size([48])
pytorch: backbone.stages.1.4.expand_conv.bn.running_var ---- torch.Size([48])
pytorch: backbone.stages.1.4.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.4.bottleneck_conv.conv.weight ---- torch.Size([48, 1, 5, 5])
pytorch: backbone.stages.1.4.bottleneck_conv.bn.weight ---- torch.Size([48])
pytorch: backbone.stages.1.4.bottleneck_conv.bn.bias ---- torch.Size([48])
pytorch: backbone.stages.1.4.bottleneck_conv.bn.running_mean ---- torch.Size([48])
pytorch: backbone.stages.1.4.bottleneck_conv.bn.running_var ---- torch.Size([48])
pytorch: backbone.stages.1.4.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.1.4.mid_se.conv1.weight ---- torch.Size([12, 48, 1, 1])
pytorch: backbone.stages.1.4.mid_se.conv1.bias ---- torch.Size([12])
pytorch: backbone.stages.1.4.mid_se.conv2.weight ---- torch.Size([48, 12, 1, 1])
pytorch: backbone.stages.1.4.mid_se.conv2.bias ---- torch.Size([48])
pytorch: backbone.stages.1.4.linear_conv.conv.weight ---- torch.Size([16, 48, 1, 1])
pytorch: backbone.stages.1.4.linear_conv.bn.weight ---- torch.Size([16])
pytorch: backbone.stages.1.4.linear_conv.bn.bias ---- torch.Size([16])
pytorch: backbone.stages.1.4.linear_conv.bn.running_mean ---- torch.Size([16])
pytorch: backbone.stages.1.4.linear_conv.bn.running_var ---- torch.Size([16])
pytorch: backbone.stages.1.4.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.0.expand_conv.conv.weight ---- torch.Size([104, 16, 1, 1])
pytorch: backbone.stages.2.0.expand_conv.bn.weight ---- torch.Size([104])
pytorch: backbone.stages.2.0.expand_conv.bn.bias ---- torch.Size([104])
pytorch: backbone.stages.2.0.expand_conv.bn.running_mean ---- torch.Size([104])
pytorch: backbone.stages.2.0.expand_conv.bn.running_var ---- torch.Size([104])
pytorch: backbone.stages.2.0.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.0.bottleneck_conv.conv.weight ---- torch.Size([104, 1, 5, 5])
pytorch: backbone.stages.2.0.bottleneck_conv.bn.weight ---- torch.Size([104])
pytorch: backbone.stages.2.0.bottleneck_conv.bn.bias ---- torch.Size([104])
pytorch: backbone.stages.2.0.bottleneck_conv.bn.running_mean ---- torch.Size([104])
pytorch: backbone.stages.2.0.bottleneck_conv.bn.running_var ---- torch.Size([104])
pytorch: backbone.stages.2.0.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.0.mid_se.conv1.weight ---- torch.Size([26, 104, 1, 1])
pytorch: backbone.stages.2.0.mid_se.conv1.bias ---- torch.Size([26])
pytorch: backbone.stages.2.0.mid_se.conv2.weight ---- torch.Size([104, 26, 1, 1])
pytorch: backbone.stages.2.0.mid_se.conv2.bias ---- torch.Size([104])
pytorch: backbone.stages.2.0.linear_conv.conv.weight ---- torch.Size([32, 104, 1, 1])
pytorch: backbone.stages.2.0.linear_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.2.0.linear_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.2.0.linear_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.2.0.linear_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.2.0.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.1.expand_conv.conv.weight ---- torch.Size([200, 32, 1, 1])
pytorch: backbone.stages.2.1.expand_conv.bn.weight ---- torch.Size([200])
pytorch: backbone.stages.2.1.expand_conv.bn.bias ---- torch.Size([200])
pytorch: backbone.stages.2.1.expand_conv.bn.running_mean ---- torch.Size([200])
pytorch: backbone.stages.2.1.expand_conv.bn.running_var ---- torch.Size([200])
pytorch: backbone.stages.2.1.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.1.bottleneck_conv.conv.weight ---- torch.Size([200, 1, 5, 5])
pytorch: backbone.stages.2.1.bottleneck_conv.bn.weight ---- torch.Size([200])
pytorch: backbone.stages.2.1.bottleneck_conv.bn.bias ---- torch.Size([200])
pytorch: backbone.stages.2.1.bottleneck_conv.bn.running_mean ---- torch.Size([200])
pytorch: backbone.stages.2.1.bottleneck_conv.bn.running_var ---- torch.Size([200])
pytorch: backbone.stages.2.1.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.1.mid_se.conv1.weight ---- torch.Size([50, 200, 1, 1])
pytorch: backbone.stages.2.1.mid_se.conv1.bias ---- torch.Size([50])
pytorch: backbone.stages.2.1.mid_se.conv2.weight ---- torch.Size([200, 50, 1, 1])
pytorch: backbone.stages.2.1.mid_se.conv2.bias ---- torch.Size([200])
pytorch: backbone.stages.2.1.linear_conv.conv.weight ---- torch.Size([32, 200, 1, 1])
pytorch: backbone.stages.2.1.linear_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.2.1.linear_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.2.1.linear_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.2.1.linear_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.2.1.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.2.expand_conv.conv.weight ---- torch.Size([200, 32, 1, 1])
pytorch: backbone.stages.2.2.expand_conv.bn.weight ---- torch.Size([200])
pytorch: backbone.stages.2.2.expand_conv.bn.bias ---- torch.Size([200])
pytorch: backbone.stages.2.2.expand_conv.bn.running_mean ---- torch.Size([200])
pytorch: backbone.stages.2.2.expand_conv.bn.running_var ---- torch.Size([200])
pytorch: backbone.stages.2.2.expand_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.2.bottleneck_conv.conv.weight ---- torch.Size([200, 1, 5, 5])
pytorch: backbone.stages.2.2.bottleneck_conv.bn.weight ---- torch.Size([200])
pytorch: backbone.stages.2.2.bottleneck_conv.bn.bias ---- torch.Size([200])
pytorch: backbone.stages.2.2.bottleneck_conv.bn.running_mean ---- torch.Size([200])
pytorch: backbone.stages.2.2.bottleneck_conv.bn.running_var ---- torch.Size([200])
pytorch: backbone.stages.2.2.bottleneck_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.2.mid_se.conv1.weight ---- torch.Size([50, 200, 1, 1])
pytorch: backbone.stages.2.2.mid_se.conv1.bias ---- torch.Size([50])
pytorch: backbone.stages.2.2.mid_se.conv2.weight ---- torch.Size([200, 50, 1, 1])
pytorch: backbone.stages.2.2.mid_se.conv2.bias ---- torch.Size([200])
pytorch: backbone.stages.2.2.linear_conv.conv.weight ---- torch.Size([32, 200, 1, 1])
pytorch: backbone.stages.2.2.linear_conv.bn.weight ---- torch.Size([32])
pytorch: backbone.stages.2.2.linear_conv.bn.bias ---- torch.Size([32])
pytorch: backbone.stages.2.2.linear_conv.bn.running_mean ---- torch.Size([32])
pytorch: backbone.stages.2.2.linear_conv.bn.running_var ---- torch.Size([32])
pytorch: backbone.stages.2.2.linear_conv.bn.num_batches_tracked ---- torch.Size([])
pytorch: backbone.stages.2.3.conv.weight ---- torch.Size([200, 32, 1, 1])
pytorch: backbone.stages.2.3.bn.weight ---- torch.Size([200])
pytorch: backbone.stages.2.3.bn.bias ---- torch.Size([200])
pytorch: backbone.stages.2.3.bn.running_mean ---- torch.Size([200])
pytorch: backbone.stages.2.3.bn.running_var ---- torch.Size([200])
pytorch: backbone.stages.2.3.bn.num_batches_tracked ---- torch.Size([])
pytorch: neck.ins_conv.0.in_conv.weight ---- torch.Size([96, 8, 1, 1])
pytorch: neck.ins_conv.0.se_block.conv1.weight ---- torch.Size([24, 96, 1, 1])
pytorch: neck.ins_conv.0.se_block.conv1.bias ---- torch.Size([24])
pytorch: neck.ins_conv.0.se_block.conv2.weight ---- torch.Size([96, 24, 1, 1])
pytorch: neck.ins_conv.0.se_block.conv2.bias ---- torch.Size([96])
pytorch: neck.ins_conv.1.in_conv.weight ---- torch.Size([96, 16, 1, 1])
pytorch: neck.ins_conv.1.se_block.conv1.weight ---- torch.Size([24, 96, 1, 1])
pytorch: neck.ins_conv.1.se_block.conv1.bias ---- torch.Size([24])
pytorch: neck.ins_conv.1.se_block.conv2.weight ---- torch.Size([96, 24, 1, 1])
pytorch: neck.ins_conv.1.se_block.conv2.bias ---- torch.Size([96])
pytorch: neck.ins_conv.2.in_conv.weight ---- torch.Size([96, 200, 1, 1])
pytorch: neck.ins_conv.2.se_block.conv1.weight ---- torch.Size([24, 96, 1, 1])
pytorch: neck.ins_conv.2.se_block.conv1.bias ---- torch.Size([24])
pytorch: neck.ins_conv.2.se_block.conv2.weight ---- torch.Size([96, 24, 1, 1])
pytorch: neck.ins_conv.2.se_block.conv2.bias ---- torch.Size([96])
pytorch: neck.inp_conv.0.in_conv.weight ---- torch.Size([24, 96, 3, 3])
pytorch: neck.inp_conv.0.se_block.conv1.weight ---- torch.Size([6, 24, 1, 1])
pytorch: neck.inp_conv.0.se_block.conv1.bias ---- torch.Size([6])
pytorch: neck.inp_conv.0.se_block.conv2.weight ---- torch.Size([24, 6, 1, 1])
pytorch: neck.inp_conv.0.se_block.conv2.bias ---- torch.Size([24])
pytorch: neck.inp_conv.1.in_conv.weight ---- torch.Size([24, 96, 3, 3])
pytorch: neck.inp_conv.1.se_block.conv1.weight ---- torch.Size([6, 24, 1, 1])
pytorch: neck.inp_conv.1.se_block.conv1.bias ---- torch.Size([6])
pytorch: neck.inp_conv.1.se_block.conv2.weight ---- torch.Size([24, 6, 1, 1])
pytorch: neck.inp_conv.1.se_block.conv2.bias ---- torch.Size([24])
pytorch: neck.inp_conv.2.in_conv.weight ---- torch.Size([24, 96, 3, 3])
pytorch: neck.inp_conv.2.se_block.conv1.weight ---- torch.Size([6, 24, 1, 1])
pytorch: neck.inp_conv.2.se_block.conv1.bias ---- torch.Size([6])
pytorch: neck.inp_conv.2.se_block.conv2.weight ---- torch.Size([24, 6, 1, 1])
pytorch: neck.inp_conv.2.se_block.conv2.bias ---- torch.Size([24])
pytorch: head.binarize.conv1.weight ---- torch.Size([24, 96, 3, 3])
pytorch: head.binarize.conv_bn1.weight ---- torch.Size([24])
pytorch: head.binarize.conv_bn1.bias ---- torch.Size([24])
pytorch: head.binarize.conv_bn1.running_mean ---- torch.Size([24])
pytorch: head.binarize.conv_bn1.running_var ---- torch.Size([24])
pytorch: head.binarize.conv_bn1.num_batches_tracked ---- torch.Size([])
pytorch: head.binarize.conv2.weight ---- torch.Size([24, 24, 2, 2])
pytorch: head.binarize.conv2.bias ---- torch.Size([24])
pytorch: head.binarize.conv_bn2.weight ---- torch.Size([24])
pytorch: head.binarize.conv_bn2.bias ---- torch.Size([24])
pytorch: head.binarize.conv_bn2.running_mean ---- torch.Size([24])
pytorch: head.binarize.conv_bn2.running_var ---- torch.Size([24])
pytorch: head.binarize.conv_bn2.num_batches_tracked ---- torch.Size([])
pytorch: head.binarize.conv3.weight ---- torch.Size([24, 1, 2, 2])
pytorch: head.binarize.conv3.bias ---- torch.Size([1])
pytorch: head.thresh.conv1.weight ---- torch.Size([24, 96, 3, 3])
pytorch: head.thresh.conv_bn1.weight ---- torch.Size([24])
pytorch: head.thresh.conv_bn1.bias ---- torch.Size([24])
pytorch: head.thresh.conv_bn1.running_mean ---- torch.Size([24])
pytorch: head.thresh.conv_bn1.running_var ---- torch.Size([24])
pytorch: head.thresh.conv_bn1.num_batches_tracked ---- torch.Size([])
pytorch: head.thresh.conv2.weight ---- torch.Size([24, 24, 2, 2])
pytorch: head.thresh.conv2.bias ---- torch.Size([24])
pytorch: head.thresh.conv_bn2.weight ---- torch.Size([24])
pytorch: head.thresh.conv_bn2.bias ---- torch.Size([24])
pytorch: head.thresh.conv_bn2.running_mean ---- torch.Size([24])
pytorch: head.thresh.conv_bn2.running_var ---- torch.Size([24])
pytorch: head.thresh.conv_bn2.num_batches_tracked ---- torch.Size([])
pytorch: head.thresh.conv3.weight ---- torch.Size([24, 1, 2, 2])
pytorch: head.thresh.conv3.bias ---- torch.Size([1])
pytorch: backbone.conv.conv.weight, torch.Size([8, 3, 3, 3])
Traceback (most recent call last):
File "./converter/ch_ppocr_v3_det_converter.py", line 51, in load_paddle_weights
self.net.state_dict()[k].copy(torch.Tensor(para_state_dict[ppname]))
KeyError: 'Student2.backbone.conv.conv.weight'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "./converter/ch_ppocr_v3_det_converter.py", line 73, in
converter = PPOCRv3DetConverter(cfg, paddle_pretrained_model_path)
File "./converter/ch_ppocr_v3_det_converter.py", line 13, in init
self.load_paddle_weights(paddle_pretrained_model_path)
File "./converter/ch_ppocr_v3_det_converter.py", line 54, in load_paddle_weights
print('paddle: {}, {}'.format(ppname, para_state_dict[ppname].shape))
KeyError: 'Student2.backbone.conv.conv.weight

[zhangshabao]

这个好像是在paddle模型的最后一层

[frotms]

转换使用是哪一个模型？名字？

[zhangshabao]

train.log
ch_PP-OCRv3_det_student.yml 这个config

[zhangshabao]

这个好像是在paddle模型的最后一层

是第一层就出现了问题

[frotms]

加qq聊下吧，42894130