Run graph2seq.py raise an error

[khaliji]

Hello, When I run this command, I get an error: python graph2seq_mp.py --dataset yelp --fanouts -1 -1 --save_dir seq_data --train_size 0.4 --val_size 0.1 --n_workers 8 --add_self_loop --norm_feat The error is: _pickle.PicklingError: Can't pickle <class 'data_utils.GraphData'>: attribute lookup GraphData on data_utils failed How can I fix it?

[Orion-wyc]

Hello, When I run this command, I get an error: python graph2seq_mp.py --dataset yelp --fanouts -1 -1 --save_dir seq_data --train_size 0.4 --val_size 0.1 --n_workers 8 --add_self_loop --norm_feat The error is: _pickle.PicklingError: Can't pickle <class 'data_utils.GraphData'>: attribute lookup GraphData on data_utils failed How can I fix it?

Could you please provide me with the python version and system environment used? It is recommended to switch to ubuntu for execution. There seems to be some problems with multiprocessing under windows. I'll look into why later.

[khaliji]

Thanks for your response Yes, I run it on Windows 10 I removed the multiprocessing code, and it worked fine now.

[MarkeyGeng]

Hi， when I use this command ‘python graph2seq_mp.py --dataset amazon --fanouts -1 -1 --save_dir seq_data --train_size 0.4 --val_size 0.1 --n_workers 8 --add_self_loop --norm_feat’ the progress bar does not change like this same error in both yelp and amazon. Could you please tell me how to solve the problem?

[Orion-wyc]

Hi， when I use this command ‘python graph2seq_mp.py --dataset amazon --fanouts -1 -1 --save_dir seq_data --train_size 0.4 --val_size 0.1 --n_workers 8 --add_self_loop --norm_feat’ the progress bar does not change like this same error in both yelp and amazon. Could you please tell me how to solve the problem?

Hi, Could you please paste your environment here? The script is only verified on linux.

[MarkeyGeng]

Hi， when I use this command ‘python graph2seq_mp.py --dataset amazon --fanouts -1 -1 --save_dir seq_data --train_size 0.4 --val_size 0.1 --n_workers 8 --add_self_loop --norm_feat’ the progress bar does not change like this same error in both yelp and amazon. Could you please tell me how to solve the problem?

Hi, Could you please paste your environment here? The script is only verified on linux.

Hi, I found that when using small-scale data, the progress bar can be displayed normally. My env: python=3.8 torch=1.12 with cuda10.2 and dgl=0.9.1

[Orion-wyc]

@MarkeyGeng Hi, Below is the environment and platform that I used for reproduction. The code can be run on both CPU and GPU platforms. Please check whether the dataset has been downloaded completely.

Linux chopin 6.4.6-76060406-generic #202307241739~1690928105~22.04~d567a38 SMP PREEMPT_DYNAMIC Tue A x86_64 x86_64 x86_64 GNU/Linux

run.log gaga_env.txt

[MarkeyGeng]

@Orion-wyc Thank you for your answer. The program is now running normally.

[zlyzly666]

hello, when I run the code, I got the result like this

how should I improve it？

[Orion-wyc]

hello, when I run the code, I got the result like this how should I improve it？

Hi, Could you please attach the running log in the comment? Judging from the pictures, it is possible that the dataset has not been downloaded completely. Try to remove ~/.dgl and run the script again.

[zlyzly666]

hello, when I run the code, I got the result like this how should I improve it？↳

Hi, Could you please attach the running log in the comment? Judging from the pictures, it is possible that the dataset has not been downloaded completely. Try to remove ~/.dgl and run the script again.↳

Thanks for your reply, below is the dataset

[Orion-wyc]

hello, when I run the code, I got the result like this how should I improve it？↳

Hi, Could you please attach the running log in the comment? Judging from the pictures, it is possible that the dataset has not been downloaded completely. Try to remove ~/.dgl and run the script again.↳

Thanks for your reply, below is the dataset

Hi， Did you follow the following steps: split the data set, preprocess the data set, and execute the training script. The training log can be saved through nohup.

If the problem is still not solved, you can package the code, dataset, and configuration files, and send it to me by email.