I have made no changes to the configs, or any other files.
The exact command I run is (copy-pasted from the page linked above):
../../tools/lazyconfig_train_net.py --config-file configs/path/to/config.py
Everything works smoothly, and the model also starts to train, but a lot of warning messages are generated, that point to different sizes of the norm.bias and norm.weight of the checkpoint and the model.
The relevant part of the output log is:
[11/04 11:35:50 fvcore.common.checkpoint]: [Checkpointer] Loading from detectron2://ImageNetPretrained/MAE/mae_pretrain_vit_base.pth ...
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_2.4.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_2.4.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_2.5.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_2.5.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_3.1.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_3.1.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_3.2.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_3.2.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_4.0.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_4.0.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_4.1.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_4.1.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_5.1.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_5.1.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of backbone.simfp_5.2.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of backbone.simfp_5.2.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv1.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv1.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv2.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv2.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv3.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv3.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv4.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.box_head.conv4.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn1.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn1.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn2.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn2.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn3.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn3.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.bias in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn4.norm.bias in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.bias will not be loaded. Please double check and see if this is desired.
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: Shape of norm.weight in checkpoint is torch.Size([768]), while shape of roi_heads.mask_head.mask_fcn4.norm.weight in model is torch.Size([256]).
WARNING [11/04 11:35:50 d2.checkpoint.c2_model_loading]: norm.weight will not be loaded. Please double check and see if this is desired.
[11/04 11:35:50 d2.checkpoint.c2_model_loading]: Following weights matched with submodule backbone.net:
| Names in Model | Names in Checkpoint | Shapes |
|:----------------------|:----------------------------------|:---------------------|
| blocks.0.attn.proj.* | blocks.0.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.0.attn.qkv.* | blocks.0.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.0.mlp.fc1.* | blocks.0.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.0.mlp.fc2.* | blocks.0.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.0.norm1.* | blocks.0.norm1.{bias,weight} | (768,) (768,) |
| blocks.0.norm2.* | blocks.0.norm2.{bias,weight} | (768,) (768,) |
| blocks.1.attn.proj.* | blocks.1.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.1.attn.qkv.* | blocks.1.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.1.mlp.fc1.* | blocks.1.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.1.mlp.fc2.* | blocks.1.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.1.norm1.* | blocks.1.norm1.{bias,weight} | (768,) (768,) |
| blocks.1.norm2.* | blocks.1.norm2.{bias,weight} | (768,) (768,) |
| blocks.10.attn.proj.* | blocks.10.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.10.attn.qkv.* | blocks.10.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.10.mlp.fc1.* | blocks.10.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.10.mlp.fc2.* | blocks.10.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.10.norm1.* | blocks.10.norm1.{bias,weight} | (768,) (768,) |
| blocks.10.norm2.* | blocks.10.norm2.{bias,weight} | (768,) (768,) |
| blocks.11.attn.proj.* | blocks.11.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.11.attn.qkv.* | blocks.11.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.11.mlp.fc1.* | blocks.11.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.11.mlp.fc2.* | blocks.11.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.11.norm1.* | blocks.11.norm1.{bias,weight} | (768,) (768,) |
| blocks.11.norm2.* | blocks.11.norm2.{bias,weight} | (768,) (768,) |
| blocks.2.attn.proj.* | blocks.2.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.2.attn.qkv.* | blocks.2.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.2.mlp.fc1.* | blocks.2.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.2.mlp.fc2.* | blocks.2.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.2.norm1.* | blocks.2.norm1.{bias,weight} | (768,) (768,) |
| blocks.2.norm2.* | blocks.2.norm2.{bias,weight} | (768,) (768,) |
| blocks.3.attn.proj.* | blocks.3.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.3.attn.qkv.* | blocks.3.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.3.mlp.fc1.* | blocks.3.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.3.mlp.fc2.* | blocks.3.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.3.norm1.* | blocks.3.norm1.{bias,weight} | (768,) (768,) |
| blocks.3.norm2.* | blocks.3.norm2.{bias,weight} | (768,) (768,) |
| blocks.4.attn.proj.* | blocks.4.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.4.attn.qkv.* | blocks.4.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.4.mlp.fc1.* | blocks.4.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.4.mlp.fc2.* | blocks.4.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.4.norm1.* | blocks.4.norm1.{bias,weight} | (768,) (768,) |
| blocks.4.norm2.* | blocks.4.norm2.{bias,weight} | (768,) (768,) |
| blocks.5.attn.proj.* | blocks.5.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.5.attn.qkv.* | blocks.5.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.5.mlp.fc1.* | blocks.5.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.5.mlp.fc2.* | blocks.5.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.5.norm1.* | blocks.5.norm1.{bias,weight} | (768,) (768,) |
| blocks.5.norm2.* | blocks.5.norm2.{bias,weight} | (768,) (768,) |
| blocks.6.attn.proj.* | blocks.6.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.6.attn.qkv.* | blocks.6.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.6.mlp.fc1.* | blocks.6.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.6.mlp.fc2.* | blocks.6.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.6.norm1.* | blocks.6.norm1.{bias,weight} | (768,) (768,) |
| blocks.6.norm2.* | blocks.6.norm2.{bias,weight} | (768,) (768,) |
| blocks.7.attn.proj.* | blocks.7.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.7.attn.qkv.* | blocks.7.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.7.mlp.fc1.* | blocks.7.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.7.mlp.fc2.* | blocks.7.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.7.norm1.* | blocks.7.norm1.{bias,weight} | (768,) (768,) |
| blocks.7.norm2.* | blocks.7.norm2.{bias,weight} | (768,) (768,) |
| blocks.8.attn.proj.* | blocks.8.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.8.attn.qkv.* | blocks.8.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.8.mlp.fc1.* | blocks.8.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.8.mlp.fc2.* | blocks.8.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.8.norm1.* | blocks.8.norm1.{bias,weight} | (768,) (768,) |
| blocks.8.norm2.* | blocks.8.norm2.{bias,weight} | (768,) (768,) |
| blocks.9.attn.proj.* | blocks.9.attn.proj.{bias,weight} | (768,) (768,768) |
| blocks.9.attn.qkv.* | blocks.9.attn.qkv.{bias,weight} | (2304,) (2304,768) |
| blocks.9.mlp.fc1.* | blocks.9.mlp.fc1.{bias,weight} | (3072,) (3072,768) |
| blocks.9.mlp.fc2.* | blocks.9.mlp.fc2.{bias,weight} | (768,) (768,3072) |
| blocks.9.norm1.* | blocks.9.norm1.{bias,weight} | (768,) (768,) |
| blocks.9.norm2.* | blocks.9.norm2.{bias,weight} | (768,) (768,) |
| patch_embed.proj.* | patch_embed.proj.{bias,weight} | (768,) (768,3,16,16) |
| pos_embed | pos_embed | (1, 197, 768) |
WARNING [11/04 11:35:50 fvcore.common.checkpoint]: Some model parameters or buffers are not found in the checkpoint:
backbone.net.blocks.0.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.1.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.10.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.11.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.2.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.3.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.4.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.5.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.6.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.7.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.8.attn.{rel_pos_h, rel_pos_w}
backbone.net.blocks.9.attn.{rel_pos_h, rel_pos_w}
backbone.simfp_2.0.{bias, weight}
backbone.simfp_2.1.{bias, weight}
backbone.simfp_2.3.{bias, weight}
backbone.simfp_2.4.norm.{bias, weight}
backbone.simfp_2.4.weight
backbone.simfp_2.5.norm.{bias, weight}
backbone.simfp_2.5.weight
backbone.simfp_3.0.{bias, weight}
backbone.simfp_3.1.norm.{bias, weight}
backbone.simfp_3.1.weight
backbone.simfp_3.2.norm.{bias, weight}
backbone.simfp_3.2.weight
backbone.simfp_4.0.norm.{bias, weight}
backbone.simfp_4.0.weight
backbone.simfp_4.1.norm.{bias, weight}
backbone.simfp_4.1.weight
backbone.simfp_5.1.norm.{bias, weight}
backbone.simfp_5.1.weight
backbone.simfp_5.2.norm.{bias, weight}
backbone.simfp_5.2.weight
proposal_generator.rpn_head.anchor_deltas.{bias, weight}
proposal_generator.rpn_head.conv.conv0.{bias, weight}
proposal_generator.rpn_head.conv.conv1.{bias, weight}
proposal_generator.rpn_head.objectness_logits.{bias, weight}
roi_heads.box_head.conv1.norm.{bias, weight}
roi_heads.box_head.conv1.weight
roi_heads.box_head.conv2.norm.{bias, weight}
roi_heads.box_head.conv2.weight
roi_heads.box_head.conv3.norm.{bias, weight}
roi_heads.box_head.conv3.weight
roi_heads.box_head.conv4.norm.{bias, weight}
roi_heads.box_head.conv4.weight
roi_heads.box_head.fc1.{bias, weight}
roi_heads.box_predictor.bbox_pred.{bias, weight}
roi_heads.box_predictor.cls_score.{bias, weight}
roi_heads.mask_head.deconv.{bias, weight}
roi_heads.mask_head.mask_fcn1.norm.{bias, weight}
roi_heads.mask_head.mask_fcn1.weight
roi_heads.mask_head.mask_fcn2.norm.{bias, weight}
roi_heads.mask_head.mask_fcn2.weight
roi_heads.mask_head.mask_fcn3.norm.{bias, weight}
roi_heads.mask_head.mask_fcn3.weight
roi_heads.mask_head.mask_fcn4.norm.{bias, weight}
roi_heads.mask_head.mask_fcn4.weight
roi_heads.mask_head.predictor.{bias, weight}
WARNING [11/04 11:35:50 fvcore.common.checkpoint]: The checkpoint state_dict contains keys that are not used by the model:
cls_token
norm.{bias, weight}
[11/04 11:35:50 d2.engine.train_loop]: Starting training from iteration 0
It goes on to train the model, but I stop it because I am not sure if the model architecture is correct or not.
Expected behavior:
Since everything is untouched after installation, shape mismatches are should not occur. I am not clear why this is happening. If there is a way to change the model settings so that these warnings do not occur, then that would be of great help. Thank you.
Environment:
Paste the output of the following command:
---------------------- -------------------------------------------------------------------------
sys.platform win32
Python 3.8.13 (default, Oct 19 2022, 22:38:03) [MSC v.1916 64 bit (AMD64)]
numpy 1.23.3
detectron2 0.6 @c:\ssl_imbalance\testproject2\detectron2\detectron2
Compiler MSVC 192930141
CUDA compiler not available
DETECTRON2_ENV_MODULE <not set>
PyTorch 1.12.1 @C:\APPS\Anaconda3\envs\TestProject2\lib\site-packages\torch
PyTorch debug build False
GPU available Yes
GPU 0,1,2 NVIDIA RTX A6000 (arch=8.6)
Driver version
CUDA_HOME None - invalid!
Pillow 9.2.0
torchvision 0.13.1 @C:\APPS\Anaconda3\envs\TestProject2\lib\site-packages\torchvision
torchvision arch flags C:\APPS\Anaconda3\envs\TestProject2\lib\site-packages\torchvision\_C.pyd
fvcore 0.1.5.post20220512
iopath 0.1.9
cv2 Not found
---------------------- -------------------------------------------------------------------------
PyTorch built with:
- C++ Version: 199711
- MSVC 192829337
- Intel(R) Math Kernel Library Version 2020.0.2 Product Build 20200624 for Intel(R) 64 architecture applications
- Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
- OpenMP 2019
- LAPACK is enabled (usually provided by MKL)
- CPU capability usage: AVX2
- CUDA Runtime 11.3
- NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_37,code=compute_37
- CuDNN 8.3.2 (built against CUDA 11.5)
- Magma 2.5.4
- Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.3, CUDNN_VERSION=8.3.2, CXX_COMPILER=C:/cb/pytorch_1000000000000/work/tmp_bin/sccache-cl.exe, CXX_FLAGS=/DWIN32 /D_WINDOWS /GR /EHsc /w /bigobj -DUSE_PTHREADPOOL -openmp:experimental -IC:/cb/pytorch_1000000000000/work/mkl/include -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.12.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=OFF, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=OFF, USE_OPENMP=ON, USE_ROCM=OFF,```
Instructions To Reproduce the Issue:
I installed Detectron2 and attempted to train the ViTDet base model from the documentation provided here: https://github.com/facebookresearch/detectron2/tree/main/projects/ViTDet
I have made no changes to the configs, or any other files.
The exact command I run is (copy-pasted from the page linked above):
../../tools/lazyconfig_train_net.py --config-file configs/path/to/config.py
Everything works smoothly, and the model also starts to train, but a lot of warning messages are generated, that point to different sizes of the
norm.bias
andnorm.weight
of the checkpoint and the model. The relevant part of the output log is:It goes on to train the model, but I stop it because I am not sure if the model architecture is correct or not.
Expected behavior:
Since everything is untouched after installation, shape mismatches are should not occur. I am not clear why this is happening. If there is a way to change the model settings so that these warnings do not occur, then that would be of great help. Thank you.
Environment:
Paste the output of the following command: