zrrskywalker / point-m2ae Goto Github PK

Here are the results of my fine-tuning:

Model parameter loading process:

The best result I got from reproducing fine-tuning:

You gave:

The model parameter loading process I got using the pre-trained model you gave:

The best result obtained from fine-tuning:

The model parameter loading process is inconsistent. Is it a problem with my pre-training?

The training step script i used is :
Pretraining got the pretarin model ckpt-best.pth
python main.py --config cfgs/pre-training/point-m2ae.yaml --exp_name pre-train

Then i copy this ckpt-best.pth to ckpts document, and use this scipt to funtune on modelnet40:
python main.py --config cfgs/fine-tuning/modelnet40.yaml --finetune_model --exp_name finetune --ckpts ckpts/ckpt-best.pth

Thanks for the super interesting work.
I am eager to experiment with your proposed method. Can you please let me know when will the codes be released?

Meets the error when loads the best ckpt to evaluate the SVM MODELNET40
python main.py --config cfgs/pre-training/point-m2ae.yaml --exp_name test_svm --test_svm modelnet40 --ckpts ./ckpts/ckpt-best.pth
RuntimeError: Error(s) in loading state_dict for Point_M2AE: Missing key(s) in state_dict: "h_encoder.token_embed.0.first_conv.0.weight", "h_encoder.token_embed.0.first_conv.0.bias", "h_encoder.token_embed.0.first_conv.1.weight", "h_encoder.token_embed.0.first_conv.1.bias", "h_encoder.token_embed.0.first_conv.1.running_mean", "h_encoder.token_embed.0.first_conv.1.running_var", "h_encoder.token_embed.0.first_conv.3.weight", "h_encoder.token_embed.0.first_conv.3.bias", "h_encoder.token_embed.0.second_conv.0.weight", "h_encoder.token_embed.0.second_conv.0.bias", "h_encoder.token_embed.0.second_conv.1.weight", "h_encoder.token_embed.0.second_conv.1.bias", "h_encoder.token_embed.0.second_conv.1.running_mean", "h_encoder.token_embed.0.second_conv.1.running_var", "h_encoder.token_embed.0.second_conv.3.weight", "h_encoder.token_embed.0.second_conv.3.bias", "h_encoder.token_embed.1.first_conv.0.weight", "h_encoder.token_embed.1.first_conv.0.bias", "h_encoder.token_embed.1.first_conv.1.weight", "h_encoder.token_embed.1.first_conv.1.bias"
, "h_encoder.token_embed.1.first_conv.1.running_mean", "h_encoder.token_embed.1.first_conv.1.running_var", "h_encoder.token_embed.1.first_conv.3.weight", "h_encoder.token_embed.1.first_conv.3.bias", "h_encoder.token_embed.1.second_conv.0.weight", "h_encoder.token_embed.1.second_conv.0.bias", "h_encoder.token_embed.1.second_conv.1.weight", "h_encoder.token_embed.1.second_conv.1.bias", "h_encoder.token_embed.1.second_conv.1.running_mean", "h_encoder.token_embed.1.second_conv.1.running_var", "h_encoder.token_embed.1.second_conv.3.weight", "h_encoder.token_embed.1.second_conv.3.bias", "h_encoder.token_embed.2.first_conv.0.weight", "h_encoder.token_embed.2.first_conv.0.bias", "h_encoder.token_embed.2.first_conv.1.weight", "h_encoder.token_embed.2.first_conv.1.bias", "h_encoder.token_embed.2.first_conv.1.running_mean", "h_encoder.token_embed.2.first_conv.1.running_var", "h_encoder.token_embed.2.first_conv.3.weight", "h_encoder.token_embed.2.first_conv.3.bias", "h_encoder.token_embed.2.second_conv.0.weight", "h_encoder.token_embed.2.second_conv.0.bias", "h_encoder.token_embed.2.second_conv.1.weight", "h_encoder.token_embed.2.second_conv.1.bias", "h_encoder.token_embed.2.second_conv.1.running_mean", "h_encoder.token_embed.2.second_conv.1.running_var", "h_encoder.token_embed.2.second_conv.3.weight", "h_encoder.token_embed.2.second_conv.3.bias", "h_encoder.encoder_pos_embeds.0.0.weight", "h_encoder.encoder_pos_embeds.0.0.bias", "h_encoder.encoder_pos_embeds.0.2.weight", "h_encoder.encoder_pos_embeds.0.2.bias", "h_encoder.encoder_pos_embeds.1.0.weight", "h_encoder.encoder_pos_embeds.1.0.bias", "h_encoder.encoder_pos_embeds.1.2.weight", "h_encoder.encoder_pos_embeds.1.2.bias", "h_encoder.encoder_pos_embeds.2.0.weight", "h_encoder.encoder_pos_embeds.2.0.bias", "h_e
ncoder.encoder_pos_embeds.2.2.weight", "h_encoder.encoder_pos_embeds.2.2.bias", "h_encoder.encoder_blocks.0.blocks.0.norm1.weight", "h_encoder.encoder_blocks.0.blocks.0.norm1.bias", "h_encoder.encoder_blocks.0.blocks.0.norm2.weight", "h_encoder.encoder_blocks.0.blocks.0.norm2.bias", "h_encoder.encoder_blocks.0.blocks.0.mlp.fc1.weight", "h_encoder.encoder_blocks.0.blocks.0.mlp.fc1.bias", "h_encoder.encoder_blocks.0.blocks.0.mlp.fc2.weight", "h_encoder.encoder_blocks.0.blocks.0.mlp.fc2.bias", "h_encoder.encoder_blocks.0.blocks.0.attn.qkv.weight", "h_encoder.encoder_blocks.0.blocks.0.attn.proj.weight", "h_encoder.encoder_blocks.0.blocks.0.attn.proj.bias", "h_encoder.encoder_blocks.0.blocks.1.norm1.weight", "h_encoder.encoder_blocks.0.blocks.1.norm1.bias", "h_encoder.encoder_blocks.0.blocks.1.norm2.weight", "h_encoder.encoder_blocks.0.blocks.1.norm2.bias", "h_encoder.encoder_blocks.0.blocks.1.mlp.fc1.weight", "h_encoder.encoder_blocks.0.blocks.1.mlp.fc1.bias", "h_encoder.encoder_blocks.0.blocks.1.mlp.fc2.weight", "h_encoder.encoder_blocks.0.blocks.1.mlp.fc2.bias", "h_encoder.encoder_blocks.0.blocks.1.attn.qkv.weight", "h_encoder.encoder_blocks.0.blocks.1.attn.proj.weight", "h_encoder.encoder_blocks.0.blocks.1.attn.proj.bias", "h_encoder.encoder_blocks.0.blocks.2.norm1.weight", "h_encoder.encoder_blocks.0.blocks.2.norm1.bias", "h_encoder.encoder_blocks.0.blocks.2.norm2.weight", "h_encoder.encoder_blocks.0.blocks.2.norm2.bias", "h_encoder.encoder_blocks.0.blocks.2.mlp.fc1.weight", "h_encoder.encoder_blocks.0.blocks.2.mlp.fc1.bias", "h_encoder.encoder_blocks.0.blocks.2.mlp.fc2.weight", "h_encoder.encoder_blocks.0.blocks.2.mlp.fc2.bias", "h_encoder.encoder_blocks.0.blocks.2.attn.qkv.weight", "h_encoder.encoder_blocks.0.blocks.2.attn.proj.weight", "h_encoder.encoder_blocks.0.blocks.2.attn.proj.bias", "h_encoder.encoder_blocks.0.blocks.3.norm1.weight", "h_encoder.encoder_blocks.0.blocks.3.norm1.bias", "h_encoder.encoder_blocks.0.blocks.3.norm2.weight", "h_encoder.encoder_blocks.0.blocks.3.norm2.bias", "h_encoder.encoder_blocks.0.blocks.3.mlp.fc1.weight", "h_encoder.encoder_blocks.0.blocks.3.mlp.fc1.bias", "h_encoder.encoder_blocks.0.blocks.3.mlp.fc2.weight", "h_encoder.encoder_blocks.0.blocks.3.mlp.fc2.bias", "h_encoder.encoder_blocks.0.blocks.3.attn.qkv.weight", "h_encoder.encoder_blocks.0.blocks.3.attn.proj.weight", "h_encoder.encoder_blocks.0.blocks.3.attn.proj.bias", "h_encoder.encoder_blocks.0.blocks.4.norm1.weight", "h_encoder.encoder_blocks.0.blocks.4.norm1.bias", "h_encoder.encoder_blocks.0.blocks.4.norm2.weight", "h_encoder.encoder_blocks.0.blocks.4.norm2.bias", "h_encoder.encoder_blocks.0.blocks.4.mlp.fc1.weight", "h_encoder.encoder_blocks.0.blocks.4.mlp.fc1.bias", "h_encoder.encoder_blocks.0.blocks.4.mlp.fc2.weight", "h_encoder.encoder_blocks.0.blocks.4.mlp.fc2.bias", "h_encoder.encoder_blocks.0.blocks.4.attn.qkv.weight", "h_encoder.encoder_blocks.0.blocks.4.attn.proj.weight", "h_encoder.encoder_blocks.0.blocks.4.attn.proj.bias", "h_encoder.encoder_blocks.1.blocks.0.norm1.weight", "h_encoder.encoder_blocks.1.blocks.0.norm1.bias", "h_encoder.encoder_blocks.1.blocks.0.norm2.weight", "h_encoder.encoder_blocks.1.blocks.0.norm2.bias", "h_encoder.encoder_blocks.1.blocks.0.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.0.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.0.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.0.mlp.fc2.bias", "h_encoder.encoder_blocks.1.blocks.0.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.0.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.0.attn.proj.bias", "h_encoder.encoder_blocks.1.blocks.1.norm1.weight", "h_encoder.encoder_blocks.1.blocks.1.norm1.bias", "h_encoder.encoder_blocks.1.blocks.1.norm2.weight", "h_encoder.encoder_blocks.1.blocks.1.norm2.bias", "h_encoder.encoder_blocks.1.blocks.1.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.1.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.1.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.1.mlp.fc2.bias", "h_encoder.encoder_blocks.1.blocks.1.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.1.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.1.attn.proj.bias", "h_encoder.encoder_blocks.1.blocks.2.norm1.weight", "h_encoder.encoder_blocks.1.blocks.2.norm1.bias", "h_encoder.encoder_blocks.1.blocks.2.norm2.weight", "h_encoder.encoder_blocks.1.blocks.2.norm2.bias", "h_encoder.encoder_blocks.1.blocks.2.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.2.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.2.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.2.mlp.fc2.bias", "h_encoder.encoder_blocks.1.blocks.2.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.2.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.2.attn.proj.bias", "h_encoder.encoder_blocks.1.blocks.3.norm1.weight", "h_encoder.encoder_blocks.1.blocks.3.norm1.bias", "h_encoder.encoder_blocks.1.blocks.3.norm2.weight", "h_encoder.encoder_blocks.1.blocks.3.norm2.bias", "h_encoder.encoder_blocks.1.blocks.3.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.3.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.3.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.3.mlp.fc2.bias", "h_encoder.encoder_blocks.1.blocks.3.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.3.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.3.attn.proj.bias", "h_encoder.encoder_blocks.1.blocks.4.norm1.weight", "h_encoder.encoder_blocks.1.blocks.4.norm1.bias", "h_encoder.encoder_blocks.1.blocks.4.norm2.weight", "h_encoder.encoder_blocks.1.blocks.4.norm2.bias", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc2.bias", "h_encoder.encoder_blocks.1.blocks.4.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.4.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.4.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.0.norm1.weightcks.4.mlp.fc1.weight", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc1.bias", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc2.weight", "h_encoder.encoder_blocks.1.blocks.4.mlp.fc2.bias", "h_encod[47/1981]r_blocks.1.blocks.4.attn.qkv.weight", "h_encoder.encoder_blocks.1.blocks.4.attn.proj.weight", "h_encoder.encoder_blocks.1.blocks.4.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.0.norm1.weight", "h_encoder.encoder_blocks.2.blocks.0.norm1.bias", "h_encoder.encoder_blocks.2.blocks.0.norm2.weight", "h_encoder.encoder_blocks.2.blocks.0.norm2.bias", "h_encoder.encoder_blocks.2.blocks.0.mlp.fc1.weight", "h_encoder.encoder_blocks.2.blocks.0.mlp.fc1.bias", "h_encoder.encoder_blocks.2.blocks.0.mlp.fc2.weight", "h_encoder.encoder_blocks.2.blocks.0.mlp.fc2.bias", "h_encoder.encoder_blocks.2.blocks.0.attn.qkv.weight", "h_encoder.encoder_blocks.2.blocks.0.attn.proj.weight", "h_encoder.encoder_blocks.2.blocks.0.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.1.norm1.weight", "h_encoder.encoder_blocks.2.blocks.1.norm1.bias", "h_encoder.encoder_blocks.2.blocks.1.norm2.weight", "h_encoder.encoder_blocks.2.blocks.1.norm2.bias", "h_encoder.encoder_blocks.2.blocks.1.mlp.fc1.weight", "h_encoder.encoder_blocks.2.blocks.1.mlp.fc1.bias", "h_encoder.encoder_blocks.2.blocks.1.mlp.fc2.weight", "h_encoder.encoder_blocks.2.blocks.1.mlp.fc2.bias", "h_encoder.encoder_blocks.2.blocks.1.attn.qkv.weight", "h_encoder.encoder_blocks.2.blocks.1.attn.proj.weight", "h_encoder.encoder_blocks.2.blocks.1.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.2.norm1.weight", "h_encoder.encoder_blocks.2.blocks.2.norm1.bias", "h_encoder.encoder_blocks.2.blocks.2.norm2.weight", "h_encoder.encoder_blocks.2.blocks.2.norm2.bias", "h_encoder.encoder_blocks.2.blocks.2.mlp.fc1.weight", "h_encoder.encoder_blocks.2.blocks.2.mlp.fc1.bias", "h_encoder.encoder_blocks.2.blocks.2.mlp.fc2.weight", "h_encoder.encoder_blocks.2.blocks.2.mlp.fc2.bias", "h_encoder.encoder_blocks.2.blocks.2.attn.qkv.weight", "h_encoder.encoder_blocks.2.blocks.2.attn.proj.weight", "h_encoder.encoder_blocks.2.blocks.2.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.3.norm1.weight", "h_encoder.encoder_blocks.2.blocks.3.norm1.bias", "h_encoder.encoder_blocks.2.blocks.3.norm2.weight", "h_encoder.encoder_blocks.2.blocks.3.norm2.bias", "h_encoder.encoder_blocks.2.blocks.3.mlp.fc1.weight", "h_encoder.encoder_blocks.2.blocks.3.mlp.fc1.bias", "h_encoder.encoder_blocks.2.blocks.3.mlp.fc2.weight", "h_encoder.encoder_blocks.2.blocks.3.mlp.fc2.bias", "h_encoder.encoder_blocks.2.blocks.3.attn.qkv.weight", "h_encoder.encoder_blocks.2.blocks.3.attn.proj.weight", "h_encoder.encoder_blocks.2.blocks.3.attn.proj.bias", "h_encoder.encoder_blocks.2.blocks.4.norm1.weight", "h_encoder.encoder_blocks.2.blocks.4.norm1.bias", "h_encoder.encoder_blocks.2.blocks.4.norm2.weight", "h_encoder.encoder_blocks.2.blocks.4.norm2.bias", "h_encoder.encoder_blocks.2.blocks.4.mlp.fc1.weight", "h_encoder.encoder_blocks.2.blocks.4.mlp.fc1.bias", "h_encoder.encoder_blocks.2.blocks.4.mlp.fc2.weight", "h_encoder.encoder_blocks.2.blocks.4.mlp.fc2.bias", "h_encoder.encoder_blocks.2.blocks.4.attn.qkv.weight", "h_encoder.encoder_blocks.2.blocks.4.attn.proj.weight", "h_encoder.encoder_blocks.2.blocks.4.attn.proj.bias", "h_encoder.encoder_norms.0.weight", "h_encoder.encoder_norms.0.bias", "h_encoder.encoder_norms.1.weight", "h_encoder.encoder_norms.1.bias", "h_encoder.encoder_norms.2.weight", "h_encoder.encoder_norms.2.bias", "h_decoder.0.blocks.0.norm1.weight", "h_decoder.0.blocks.0.norm1.bias", "h_decoder.0.blocks.0.norm2.weight", "h_decoder.0.blocks.0.norm2.bias", "h_decoder.0.blocks.0.mlp.fc1.weight", "h_decoder.0.blocks.0.mlp.fc1.bias", "h_decoder.0.blocks.0.mlp.fc2.weight", "h_decoder.0.blocks.0.mlp.fc2.bias", "h_decoder.0.blocks.0.attn.qkv.weight", "h_decoder.0.blocks.0.attn.proj.weight", "h_decoder.0.blocks.0.attn.proj.bias", "h_decoder.1.blocks.0.norm1.weight", "h_decoder.1.blocks.0.norm1.bias", "h_decoder.1.blocks.0.norm2.weight", "h_decoder.1.blocks.0.norm2.bias", "h_decoder.1.blocks.0.mlp.fc1.weight", "h_decoder.1.blocks.0.mlp.fc1.bias", "h_decoder.1.blocks.0.mlp.fc2.weight", "h_decoder.1.blocks.0.mlp.fc2.bias", "h_decoder.1.blocks.0.attn.qkv.weight", "h_decoder.1.blocks.0.attn.proj.weight", "h_decoder.1.blocks.0.attn.proj.bias", "token_prop.0.fuse.net.0.weight", "token_prop.0.fuse.net.0.bias", "token_prop.0.fuse.net.1.weight", "token_prop.0.fuse.net.1.bias", "token_prop.0.fuse.net.1.running_mean", "token_prop.0.fuse.net.1.running_var", "token_prop.0.extraction.operation.0.net1.0.weight", "token_prop.0.extraction.operation.0.net1.0.bias", "token_prop.0.extraction.operation.0.net1.1.weight", "token_prop.0.extraction.operation.0.net1.1.bias", "token_prop.0.extraction.operation.0.net1.1.running_mean", "token_prop.0.extraction.operation.0.net1.1.running_var", "token_prop.0.extraction.operation.0.net2.0.weight", "token_prop.0.extraction.operation.0.net2.0.bias", "token_prop.0.extraction.operation.0.net2.1.weight", "token_prop.0.extraction.operation.0.net2.1.bias", "token_prop.0.extraction.operation.0.net2.1.running_mean", "token_prop.0.extraction.operation.0.net2.1.running_var", "rec_head.weight", "rec_head.bias". Unexpected key(s) in state_dict: "MAE_encoder.encoders.0.first_conv.0.weight", "MAE_encoder.encoders.0.first_conv.0.bias", "MAE_encoder.encoders.0.first_conv.1.weight", "MAE_encoder.encoders.0.first_conv.1.bias", "MAE_encoder.encoders.0.first_conv.1.running_mean", "MAE_encoder.encoders.0.first_conv.1.running_var", "MAE_encoder.encoders.0.first_conv.1.num_batches_tracked", "MAE_encoder.encoders.0.first_conv.3.weight", "MAE_encoder.encoders.0.first_conv.3.bias", "MAE_encoder.encoders.0.second_conv.0.weight", "MAE_encoder.encoders.0.second_conv.0.bias", "MAE_encoder.encoders.0.second_conv.1.weight", "MAE_encoder.encoders.0.second_conv.1.bias", "MAE_encoder.encoders.0.second_conv.1.running_mean", "MAE_encoder.encoders.0.second_conv.1.running_var", "MAE_encoder.encoders.0.second_conv.1.num_batches_tracked", "MAE_encoder.encoders.0.second_conv.3.weight", "MAE_encoder.encoders.0.second_conv.3.bias", "MAE_encoder.encoders.1.first_conv.0.weight", "MAE_encoder.encoders.1.first_conv.0.bias", "MAE_encoder.encoders.1.first_conv.1.weight", "MAE_encoder.encoders.1.first_conv.1.bias", "MAE_encoder.encoders.1.first_conv.1.running_mean", "MAE_encoder.encoders.1.first_conv.1.running_var", "MAE_encoder.encoders.1.first_conv.1.num_batches_tracked", "MAE_encoder.encoders.1.first_conv.3.weight", "MAE_encoder.encoders.1.first_conv.3.bias", "MAE_encoder.encoders.1.second_conv.0.weight", "MAE_encoder.encoders.1.second_conv.0.bias", "MAE_encoder.encoders.1.second_conv.1.weight", "MAE_encoder.encoders.1.second_conv.1.bias", "MAE_encoder.encoders.1.second_conv.1.running_mean", "MAE_encoder.encoders.1.second_conv.1.running_var", "MAE_encoder.encoders.1.second_conv.1.num_batches_tracked", "MAE_encoder.encoders.1.second_conv.3.weight", "MAE_encoder.encoders.1.second_conv.3.bias", "MAE_encoder.encoders.2.first_conv.0.weight", "MAE_encoder.encoders.2.first_conv.0.bias", "MAE_encoder.encoders.2.first_conv.1.weight", "MAE_encoder.encoders.2.first_conv.1.bias", "MAE_encoder.encoders.2.first_conv.1.running_mean", "MAE_encoder.encoders.2.first_conv.1.running_var", "MAE_encoder.encoders.2.first_conv.1.num_batches_tracked", "MAE_encoder.encoders.2.first_conv.3.weight", "MAE_encoder.encoders.2.first_conv.3.bias", "MAE_encoder.encoders.2.second_conv.0.weight", "MAE_encoder.encoders.2.second_conv.0.bias", "MAE_encoder.encoders.2.second_conv.1.weight", "MAE_encoder.encoders.2.second_conv.1.bias", "MAE_encoder.encoders.2.second_conv.1.running_mean", "MAE_encoder.encoders.2.second_conv.1.running_var", "MAE_encoder.encoders.2.second_conv.1.num_batches_tracked", "MAE_encoder.encoders.2.second_conv.3.weight", "MAE_encoder.encoders.2.second_conv.3.bias", "MAE_encoder.pos_embeds.0.0.weight", "MAE_encoder.pos_embeds.0.0.bias", "MAE_encoder.pos_embeds.0.2.weight", "MAE_encoder.pos_embeds.0.2.bias", "MAE_encoder.pos_embeds.1.0.weight", "MAE_encoder.pos_embeds.1.0.bias", "MAE_encoder.pos_embeds.1.2.weight", "MAE_encoder.pos_embeds.1.2.bias", "MAE_encoder.pos_embeds.2.0.weight", "MAE_encoder.pos_embeds.2.0.bias", "MAE_encoder.pos_embeds.2.2.weight", "MAE_encoder.pos_embeds.2.2.bias", "MAE_encoder.blocks.0.blocks.0.norm1.weight", "MAE_encoder.blocks.0.blocks.0.norm1.bias", "MAE_encoder.blocks.0.blocks.0.norm2.weight", "MAE_encoder.blocks.0.blocks.0.norm2.bias", "MAE_encoder.blocks.0.blocks.0.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.0.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.0.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.0.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.0.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.0.attn.proj.weight", "MAE_encoder.blocks.0.blocks.0.attn.proj.bias", "MAE_encoder.blocks.0.blocks.1.norm1.weight", "MAE_encoder.blocks.0.blocks.1.norm1.bias", "MAE_encoder.blocks.0.blocks.1.norm2.weight", "MAE_encoder.blocks.0.blocks.1.norm2.bias", "MAE_encoder.blocks.0.blocks.1.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.1.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.1.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.1.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.1.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.1.attn.proj.weight", "MAE_encoder.blocks.0.blocks.1.attn.proj.bias", "MAE_encoder.blocks.0.blocks.2.norm1.weight", "MAE_encoder.blocks.0.blocks.2.norm1.bias", "MAE_encoder.blocks.0.blocks.2.norm2.weight", "MAE_encoder.blocks.0.blocks.2.norm2.bias", "MAE_encoder.blocks.0.blocks.2.mlp.fc1.weight", "MAE_.0.blocks.1.norm2.bias", "MAE_encoder.blocks.0.blocks.1.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.1.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.1.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.1.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.1.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.1.attn.proj.weight", "MAE_encoder.blocks.0.blocks.1.attn.proj.bias", "MAE_encoder.blocks.0.blocks.2.norm1.weight", "MAE_encoder.blocks.0.blocks.2.norm1.bias", "MAE_encoder.blocks.0.blocks.2.norm2.weight", "MAE_encoder.blocks.0.blocks.2.norm2.bias", "MAE_encoder.blocks.0.blocks.2.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.2.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.2.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.2.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.2.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.2.attn.proj.weight", "MAE_encoder.blocks.0.blocks.2.attn.proj.bias", "MAE_encoder.blocks.0.blocks.3.norm1.weight", "MAE_encoder.blocks.0.blocks.3.norm1.bias", "MAE_encoder.blocks.0.blocks.3.norm2.weight", "MAE_encoder.blocks.0.blocks.3.norm2.bias", "MAE_encoder.blocks.0.blocks.3.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.3.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.3.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.3.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.3.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.3.attn.proj.weight", "MAE_encoder.blocks.0.blocks.3.attn.proj.bias", "MAE_encoder.blocks.0.blocks.4.norm1.weight", "MAE_encoder.blocks.0.blocks.4.norm1.bias", "MAE_encoder.blocks.0.blocks.4.norm2.weight", "MAE_encoder.blocks.0.blocks.4.norm2.bias", "MAE_encoder.blocks.0.blocks.4.mlp.fc1.weight", "MAE_encoder.blocks.0.blocks.4.mlp.fc1.bias", "MAE_encoder.blocks.0.blocks.4.mlp.fc2.weight", "MAE_encoder.blocks.0.blocks.4.mlp.fc2.bias", "MAE_encoder.blocks.0.blocks.4.attn.qkv.weight", "MAE_encoder.blocks.0.blocks.4.attn.proj.weight", "MAE_encoder.blocks.0.blocks.4.attn.proj.bias", "MAE_encoder.blocks.1.blocks.0.norm1.weight", "MAE_encoder.blocks.1.blocks.0.norm1.bias", "MAE_encoder.blocks.1.blocks.0.norm2.weight", "MAE_encoder.blocks.1.blocks.0.norm2.bias", "MAE_encoder.blocks.1.blocks.0.mlp.fc1.weight", "MAE_encoder.blocks.1.blocks.0.mlp.fc1.bias", "MAE_encoder.blocks.1.blocks.0.mlp.fc2.weight", "MAE_encoder.blocks.1.blocks.0.mlp.fc2.bias", "MAE_encoder.blocks.1.blocks.0.attn.qkv.weight", "MAE_encoder.blocks.1.blocks.0.attn.proj.weight", "MAE_encoder.blocks.1.blocks.0.attn.proj.bias", "MAE_encoder.blocks.1.blocks.1.norm1.weight", "MAE_encoder.blocks.1.blocks.1.norm1.bias", "MAE_encoder.blocks.1.blocks.1.norm2.weight", "MAE_encoder.blocks.1.blocks.1.norm2.bias", "MAE_encoder.blocks.1.blocks.1.mlp.fc1.weight", "MAE_encoder.blocks.1.blocks.1.mlp.fc1.bias", "MAE_encoder.blocks.1.blocks.1.mlp.fc2.weight", "MAE_encoder.blocks.1.blocks.1.mlp.fc2.bias", "MAE_encoder.blocks.1.blocks.1.attn.qkv.weight", "MAE_encoder.blocks.1.blocks.1.attn.proj.weight", "MAE_encoder.blocks.1.blocks.1.attn.proj.bias", "MAE_encoder.blocks.1.blocks.2.norm1.weight", "MAE_encoder.blocks.1.blocks.2.norm1.bias", "MAE_encoder.blocks.1.blocks.2.norm2.weight", "MAE_encoder.blocks.1.blocks.2.norm2.bias", "MAE_encoder.blocks.1.blocks.2.mlp.fc1.weight", "MAE_encoder.blocks.1.blocks.2.mlp.fc1.bias", "MAE_encoder.blocks.1.blocks.2.mlp.fc2.weight", "MAE_encoder.blocks.1.blocks.2.mlp.fc2.bias", "MAE_encoder.blocks.1.blocks.2.attn.qkv.weight", "MAE_encoder.blocks.1.blocks.2.attn.proj.weight", "MAE_encoder.blocks.1.blocks.2.attn.proj.bias", "MAE_encoder.blocks.1.blocks.3.norm1.weight", "MAE_encoder.blocks.1.blocks.3.norm1.bias", "MAE_encoder.blocks.1.blocks.3.norm2.weight", "MAE_encoder.blocks.1.blocks.3.norm2.bias", "MAE_encoder.blocks.1.blocks.3.mlp.fc1.weight", "MAE_encoder.blocks.1.blocks.3.mlp.fc1.bias", "MAE_encoder.blocks.1.blocks.3.mlp.fc2.weight", "MAE_encoder.blocks.1.blocks.3.mlp.fc2.bias", "MAE_encoder.blocks.1.blocks.3.attn.qkv.weight", "MAE_encoder.blocks.1.blocks.3.attn.proj.weight", "MAE_encoder.blocks.1.blocks.3.attn.proj.bias", "MAE_encoder.blocks.1.blocks.4.norm1.weight", "MAE_encoder.blocks.1.blocks.4.norm1.bias", "MAE_encoder.blocks.1.blocks.4.norm2.weight", "MAE_encoder.blocks.1.blocks.4.norm2.bias", "MAE_encoder.blocks.1.blocks.4.mlp.fc1.weight", "MAE_encoder.blocks.1.blocks.4.mlp.fc1.bias", "MAE_encoder.blocks.1.blocks.4.mlp.fc2.weight", "MAE_encoder.blocks.1.blocks.4.mlp.fc2.bias", "MAE_encoder.blocks.1.blocks.4.attn.qkv.weight", "MAE_encoder.blocks.1.blocks.4.attn.proj.weight", "MAE_encoder.blocks.1.blocks.4.attn.proj.bias", "MAE_encoder.blocks.2.blocks.0.norm1.weight", "MAE_encoder.blocks.2.blocks.0.norm1.bias", "MAE_encoder.blocks.2.blocks.0.norm2.weight", "MAE_encoder.blocks.2.blocks.0.norm2.bias", "MAE_encoder.blocks.2.blocks.0.mlp.fc1.weight", "MAE_encoder.blocks.2.blocks.0.mlp.fc1.bias", "MAE_encoder.blocks.2.blocks.0.mlp.fc2.weight", "MAE_encoder.blocks.2.blocks.0.mlp.fc2.bias", "MAE_encoder.blocks.2.blocks.0.attn.qkv.weight", "MAE_encoder.blocks.2.blocks.0.attn.proj.weight", "MAE_encoder.blocks.2.blocks.0.attn.proj.bias", "MAE_encoder.blocks.2.blocks.1.norm1.weight", "MAE_encoder.blocks.2.blocks.1.norm1.bias", "MAE_encoder.blocks.2.blocks.1.norm2.weight", "MAE_encoder.blocks.2.blocks.1.norm2.bias", "MAE_encoder.blocks.2.blocks.1.mlp.fc1.weight", "MAE_encoder.blocks.2.blocks.1.mlp.fc1.bias", "MAE_encoder.blocks.2.blocks.1.mlp.fc2.weight", "MAE_encoder.blocks.2.blocks.1.mlp.fc2.bias", "MAE_encoder.blocks.2.blocks.1.attn.qkv.weight", "MAE_encoder.blocks.2.blocks.1.attn.proj.weight", "MAE_encoder.blocks.2.blocks.1.attn.proj.bias", "MAE_encoder.blocks.2.blocks.2.norm1.weight", "MAE_encoder.blocks.2.blocks.2.norm1.bias", "MAE_encoder.blocks.2.blocks.2.norm2.weight", "MAE_encoder.blocks.2.blocks.2.norm2.bias", "MAE_encoder.blocks.2.blocks.2.mlp.fc1.weight", "MAE_encoder.blocks.2.blocks.2.mlp.fc1.bias", "MAE_encoder.blocks.2.blocks.2.mlp.fc2.weight", "MAE_encoder.blocks.2.blocks.2.mlp.fc2.bias", "MAE_encoder.blocks.2.blocks.2.attn.qkv.weight", "MAE_encoder.blocks.2.blocks.2.attn.proj.weight", "MAE_encoder.blocks.2.blocks.2.attn.proj.bias", "MAE_encoder.blocks.2.blocks.3.norm1.weight", "MAE_encoder.blocks.2.blocks.3.norm1.bias", "MAE_encoder.blocks.2.blocks.3.norm2.weight", "MAE_encoder.blocks.2.blocks.3.norm2.bias", "MAE_encoder.blocks.2.blocks.3.mlp.fc1.weight", "MAE_encoder.blocks.2.blocks.3.mlp.fc1.bias", "MAE_encoder.blocks.2.blocks.3.mlp.fc2.weight", "MAE_encoder.blocks.2.blocks.3.mlp.fc2.bias", "MAE_encoder.blocks.2.blocks.3.attn.qkv.weight", "MAE_encoder.blocks.2.blocks.3.attn.proj.weight", "MAE_encoder.blocks.2.blocks.3.attn.proj.bias", "MAE_encoder.blocks.2.blocks.4.norm1.weight", "MAE_encoder.blocks.2.blocks.4.norm1.bias", "MAE_encoder.blocks.2.blocks.4.norm2.weight", "MAE_encoder.blocks.2.blocks.4.norm2.bias", "MAE_encoder.blocks.2.blocks.4.mlp.fc1.weight", "MAE_encoder.blocks.2.blocks.4.mlp.fc1.bias", "MAE_encoder.blocks.2.blocks.4.mlp.fc2.weight", "MAE_encoder.blocks.2.blocks.4.mlp.fc2.bias", "MAE_encoder.blocks.2.blocks.4.attn.qkv.weight", "MAE_encoder.blocks.2.blocks.4.attn.proj.weight", "MAE_encoder.blocks.2.blocks.4.attn.proj.bias", "MAE_encoder.norms.0.weight", "MAE_encoder.norms.0.bias", "MAE_encoder.norms.1.weight", "MAE_encoder.norms.1.bias", "MAE_encoder.norms.2.weight", "MAE_encoder.norms.2.bias", "MAE_decoders.0.blocks.0.norm1.weight", "MAE_decoders.0.blocks.0.norm1.bias", "MAE_decoders.0.blocks.0.norm2.weight", "MAE_decoders.0.blocks.0.norm2.bias", "MAE_decoders.0.blocks.0.mlp.fc1.weight", "MAE_decoders.0.blocks.0.mlp.fc1.bias", "MAE_decoders.0.blocks.0.mlp.fc2.weight", "MAE_decoders.0.blocks.0.mlp.fc2.bias", "MAE_decoders.0.blocks.0.attn.qkv.weight", "MAE_decoders.0.blocks.0.attn.proj.weight", "MAE_decoders.0.blocks.0.attn.proj.bias", "MAE_decoders.1.blocks.0.norm1.weight", "MAE_decoders.1.blocks.0.norm1.bias", "MAE_decoders.1.blocks.0.norm2.weight", "MAE_decoders.1.blocks.0.norm2.bias", "MAE_decoders.1.blocks.0.mlp.fc1.weight", "MAE_decoders.1.blocks.0.mlp.fc1.bias", "MAE_decoders.1.blocks.0.mlp.fc2.weight", "MAE_decoders.1.blocks.0.mlp.fc2.bias", "MAE_decoders.1.blocks.0.attn.qkv.weight", "MAE_decoders.1.blocks.0.attn.proj.weight", "MAE_decoders.1.blocks.0.attn.proj.bias", "up_decoders.0.fuse.net.0.weight", "up_decoders.0.fuse.net.0.bias", "up_decoders.0.fuse.net.1.weight", "up_decoders.0.fuse.net.1.bias", "up_decoders.0.fuse.net.1.running_mean", "up_decoders.0.fuse.net.1.running_var", "up_decoders.0.fuse.net.1.num_batches_tracked", "up_decoders.0.extraction.operation.0.net1.0.weight", "up_decoders.0.extraction.operation.0.net1.0.bias", "up_decoders.0.extraction.operation.0.net1.1.weight", "up_decoders.0.extraction.operation.0.net1.1.bias", "up_decoders.0.extraction.operation.0.net1.1.running_mean", "up_decoders.0.extraction.operation.0.net1.1.running_var", "up_decoders.0.extraction.operation.0.net1.1.num_batches_tracked", "up_decoders.0.extraction.operation.0.net2.0.weight", "up_decoders.0.extraction.operation.0.net2.0.bias", "up_decoders.0.extraction.operation.0.net2.1.weight", "up_decoders.0.extraction.operation.0.net2.1.bias", "up_decoders.0.extraction.operation.0.net2.1.running_mean", "up_decoders.0.extraction.operation.0.net2.1.running_var", "up_decoders.0.extraction.operation.0.net2.1.num_batches_tracked", "increase_dim.0.weight", "increase_dim.0.bias".

I'm wondering your pytorch version.
My torch vision is 2.0.0. It seems that this version will have problem about the complication of the emd_kernal.cu because of the disappearance of the THC/THC.h.
Looking forward to your reply.

Thinks

Hi, I have been trying to run your code, and I have found that pip install -r requirements.txt installs the latest Open3D package form PyPI and also installs torch v 1.13.0 from PyPI as a dependency.

conda install pytorch torchvision cudatoolkit also installs PyTorch, but the package is called pytorch and not torch. So there are two versions of PyTorch installed simultaneously in the environment.

This causes issues while executing the instruction import chamfer. Furthermore, he GCC version needs to be <=10 to successfully compile the extensions.

Can you update the README to reflect this? It would be great if you could provide your conda environment.yml file with which you were successfully able to run your code.

Hi! Thank you for your great work.

I want to know when will you release the code?

Hi! Thank you for your great work.

I want to know when will you release the code?

The dataset downloaded from the GoogleDrive link only contains the folder ShapeNet55/shapenet_pc . There are no train.txt / test.txt files. Can it be that the link is wrong?

Where are the corresponding codes of `Skip Connections'?
Thank you!

Very Great job, I would like to know more about your work, but I couldn't find the code to visualize the point cloud reconstruction of your model. Can you provide the code to visualize the reconstruction of the point cloud effect? I would greatly appreciate it!!!

I use the pre-trained model to obtain the features, but I want to obtain the per-pixel features, how can I do it?

Hello !
Do you have pertained models?

Thank you for your intriguing effort.

Can you inform me of the date when the fine-tuning code for segmentation
will be made available?