Hi, thanks for the question. My fast guess is that you overwrite an inner attribute of the WildlifeDataset class (we will try to raise warning when something like this happens in the next version of the package). If you change the next-to-last line in your code by
= ds_train_wild.get_subset(idx_train)\nwill your code work? If not, I will take a deeper look into it. I would also like to ask how is it to work with wildlife-datasets? Is there anything which was difficult to understand or which you would like to be improved?
\nBest,
\nLukas
", "rawMarkdown": "Hi, thanks for the question. My fast guess is that you overwrite an inner attribute of the `WildlifeDataset` class (we will try to raise warning when something like this happens in the next version of the package). If you change the next-to-last line in your code by\n\n```\nds_train_wild = ds_train_wild.get_subset(idx_train)\n```\n\nwill your code work? If not, I will take a deeper look into it. I would also like to ask how is it to work with wildlife-datasets? Is there anything which was difficult to understand or which you would like to be improved?\n\nBest,\n\nLukas", "votes": null }, { "id": "3165721", "postDate": "04/01/2025 00:58:45", "content": "@wildlifedatasets Thank you very much for your help, I managed to do it here. It's been a really good challenge, it caught my attention when I saw it here on Kaggle. I usually do challenges with tabular data. This one with images is being fun and I'm able to learn a lot, I'm reading your papers calmly to be able to better understand the libraries that are being used in the examples. I'm dividing the notebook of the example to understand each step. The architecture that you set up for loading images, I found a difference, very cool. Congratulations on your work.
\nThe difficulty I'm having is more in the model, as I have the practice of creating the architecture, using something more ready-made than huggingface and improving the result ends up being a challenge. Thank you very much for this opportunity, it's great to always be learning more.
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F13034312%2Fdae88b4434d1dcb9e5ecbad0208e99df%2FScreenshot%20from%202025-03-31%2021-52-22.png?generation=1743468778686286&alt=media\")
@sc0v1n0 Thanks a lot for the feedback, it is nice to know that people appreciate our work :)
", "rawMarkdown": "sc0v1n0 Thanks a lot for the feedback, it is nice to know that people appreciate our work :)", "votes": null } ], "comments": [ { "id": 3163825, "author_name": "wildlifedatasets", "author_url": "", "post_date": "03/31/2025 07:55:56", "content": "Hi, thanks for the question. My fast guess is that you overwrite an inner attribute of the WildlifeDataset class (we will try to raise warning when something like this happens in the next version of the package). If you change the next-to-last line in your code by
= ds_train_wild.get_subset(idx_train)\nwill your code work? If not, I will take a deeper look into it. I would also like to ask how is it to work with wildlife-datasets? Is there anything which was difficult to understand or which you would like to be improved?
\nBest,
\nLukas
", "votes": null, "replies": [ { "id": 3165721, "author_name": "sc0v1n0", "author_url": "", "post_date": "04/01/2025 00:58:45", "content": "@wildlifedatasets Thank you very much for your help, I managed to do it here. It's been a really good challenge, it caught my attention when I saw it here on Kaggle. I usually do challenges with tabular data. This one with images is being fun and I'm able to learn a lot, I'm reading your papers calmly to be able to better understand the libraries that are being used in the examples. I'm dividing the notebook of the example to understand each step. The architecture that you set up for loading images, I found a difference, very cool. Congratulations on your work.
\nThe difficulty I'm having is more in the model, as I have the practice of creating the architecture, using something more ready-made than huggingface and improving the result ends up being a challenge. Thank you very much for this opportunity, it's great to always be learning more.
@sc0v1n0 Thanks a lot for the feedback, it is nice to know that people appreciate our work :)
", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3163441": "I'm trying to split the set, so I can run the model faster. I've tried a few ways but when I get to extractor() I get the error below.\n\n\n\n```python\nds_train_wild = AnimalCLEF2025(INPUT_DIR, transform=transform_display, load_label=True)\nds_valid_wild = AnimalCLEF2025(INPUT_DIR, transform=transform_display, load_label=True)\n\nfrom wildlife_datasets import splits\n\nsplitter = splits.ClosedSetSplit(0.2)\nidx_train, idx_valid = splitter.split(ds_train_wild.df)[0]\n\n# df_train = ds_train_wild.df.loc[idx_train]\n\nds_train_wild.df = ds_train_wild.df.loc[idx_train]\nds_train_wild\n\n```\n\n\n```txt\n\nTypeError Traceback (most recent call last)\n