The CSV files of the training set and test set for the patent classification task.

[jali1999]

Hello, the work you have done is excellent. 😊I am currently learning IMPACT. Could you please provide the training and test set CSV files for the patent classification task?
Additionally, I have a question. I downloaded the "Patent Grant Full Text Data with Embedded TIFF Images (JAN 2001 - PRESENT) (Grant Red Book based on WIPO ST.36)" data mentioned in your Data processing description, but after processing it with folder_organize.py, I did not obtain the corresponding design patent files. Is it because I did not download the correct data?😥
Looking forward to your reply! Thank you!😁

[hhshomee]

Hi! 😊
Thanks so much for your interest in IMPACT! I’ve just uploaded the training and test CSV files for the classification task. You can now find them in the Tasks/Classification folder in the repo.

Regarding the data processing issue: please make sure you’re following the exact steps outlined in the README under the data processing section. Specifically, check if your folder structure after running the scripts matches the expected hierarchy shown there. If the hierarchy differs or you're still not seeing the DESIGN folder after folder_organize.py, feel free to share a snapshot of your extracted folder structure. It will help me identify what might be going wrong.

[jali1999]

Thank you for your reply! ☺️
But there seem to be some problems: the CSV file you just uploaded does not contain the column names of the three embedding vectors 'image_embeddings', 'embeddings_caption', 'embeddings_title', which correspond to line 16 in the classification.py file. 😮
In addition, the problem of not finding the DESIGN file after decompression has been found. The I20250626.tar compressed file I downloaded before did not contain the DESIGN file. After downloading the I20220104.tar file mentioned in the README document, I found the DESIGN file included. Looking forward to your reply again!😆