markin-wang / clevit Goto Github PK

Hi, I appreciate your work very much. Can you open source the attention visualization part of the code?

作者您好：
复现代码的过程中，遇到了一些问题，希望能得到您的解答。main.py文件里面的正样本对构建有出入，正负样本是一样的图片吗？并且data_loader_train在dataloader那边是可以取到数据的，然后到了训练的时候，图像数据就是空的，标签数据还在。
并报错

拿data_loader_val的数据集去train，虽然跑不通，后面有维度问题，但是标签和这个img里面的内容是有的。具体的维度问题是：正常val时，数据维度是[B,3,448,448]，但拿val的数据train时，经过这段代码

维度会变为[B*3,448,448]。期待您的回复，谢谢！

: )

Hello, when I run sh scripts/run_gene.sh, when loading the pre-trained model, the following error occurs:

traceback : Traceback (most recent call last):
File "/home/image/anaconda3/envs/plant_diseases/lib/python3.9/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 345, in wrapper
return f(*args, **kwargs)
File "/home/image/plant_disease/CLEViT/main.py", line 94, in main
load_pretrained(config, model_without_ddp, logger)
File "/home/image/plant_disease/CLEViT/utils.py", line 206, in load_pretrained
relative_position_bias_table_current = model.state_dict()[k]
KeyError: 'layers.2.blocks.6.attn.relative_position_bias_table'

感谢你做出的工作，我是一个初学者，我需要如何进行对比学习预训练，如何用在我自己的细粒度视觉分类数据集上呢？是修改main.py中的参数is_pretrain吗？

Hi Author, We are having the following problem with the reproduction process.

scripts/run_gene.sh: line 2: logs/gene/gene_sb7_fullb_np2_lr2e-3_mul5_bs12_448_ep200_linear_ee2: No such file or directory
tail: cannot open 'logs/gene/gene_sb7_fullb_np2_lr2e-3_mul5_bs12_448_ep200_linear_ee2' for reading: No such file or directory
tail: no files remaining