![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7279405%2F10c3326cc8d3eb7d6fa1e7d6e3cd491c%2FScreenshot%202025-04-06%20204328.png?generation=1743961471114873&alt=media\")
I accidentally deleted my data models to make room for new tasks, forgetting that the competition was not over yet. As a result, all my notebooks turned into pumpkins. And I was eliminated from the competition for this reason. I am very upset and sad, but it is my fault.
\nI apologize to those who used my datasets from a public notebook. This happened to me for the first time. Next time, I will not repeat this. I will remember my mistake for a long time.
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7279405%2F2ee31b8ec5870b8210a2d3f17395a56a%2FScreenshot%202025-04-06%20203835.png?generation=1743961155297406&alt=media\")
I accidentally deleted my data models to make room for new tasks, forgetting that the competition was not over yet. As a result, all my notebooks turned into pumpkins. And I was eliminated from the competition for this reason. I am very upset and sad, but it is my fault.
\nI apologize to those who used my datasets from a public notebook. This happened to me for the first time. Next time, I will not repeat this. I will remember my mistake for a long time.
\n
Hi, Konstantin. Everything happens for the first time, and you shouldn't blame yourself for mistakes (´-ω-`)
\nBut I'm burning with curiosity about how much \"weight' the leaderboard will lose))) (i also got an error, i used these deleted datasets)
", "rawMarkdown": "Hi, Konstantin. Everything happens for the first time, and you shouldn't blame yourself for mistakes (´-ω-`)\n\nBut I'm burning with curiosity about how much \"weight' the leaderboard will lose))) (i also got an error, i used these deleted datasets)", "votes": null }, { "id": "3173122", "postDate": "04/07/2025 14:42:58", "content": "Hi @hertryggr. In theory, it has lost weight now, I am no longer on the list, and therefore, perhaps, someone else is, if my models were used in both versions and there was no try … except block.
\nI was so busy with the AIMO-2 competition, I sat until three in the morning and I didn’t have enough space for models and datasets, and there were models from 20 to 60 GB, constant quantization of models, I eventually filled both data warehouses. The most interesting thing is that I thought that the competition was already over and I didn’t even get bronze. Sometimes such glitches happen in the head. Well, accordingly, I smashed all the models. And then I saw on the forum that you can view a hidden submission bookmark. And when I looked, I was shocked. There was absolutely no working notebook that was without errors. I tried a bunch of models from github to get something good. I created 225 versions of notebooks in experiments. I really spent 2 months on this. And I also deleted all these versions from the local computer, since LLM models filled up 2 TB of SSD. (The non-quantized LLM model takes up to 160 GB of space.) In other words, I don't have any models from this contest on the site or on my local computer. Only the notebooks remain.
\nNow I understand that I was right, I definitely won’t even get bronze now.
", "rawMarkdown": "Hi @hertryggr. In theory, it has lost weight now, I am no longer on the list, and therefore, perhaps, someone else is, if my models were used in both versions and there was no try ... except block. \n\nI was so busy with the AIMO-2 competition, I sat until three in the morning and I didn’t have enough space for models and datasets, and there were models from 20 to 60 GB, constant quantization of models, I eventually filled both data warehouses. The most interesting thing is that I thought that the competition was already over and I didn’t even get bronze. Sometimes such glitches happen in the head. Well, accordingly, I smashed all the models. And then I saw on the forum that you can view a hidden submission bookmark. And when I looked, I was shocked. There was absolutely no working notebook that was without errors. I tried a bunch of models from github to get something good. I created 225 versions of notebooks in experiments. I really spent 2 months on this. And I also deleted all these versions from the local computer, since LLM models filled up 2 TB of SSD. (The non-quantized LLM model takes up to 160 GB of space.) In other words, I don't have any models from this contest on the site or on my local computer. Only the notebooks remain.\n\nNow I understand that I was right, I definitely won’t even get bronze now.", "votes": null }, { "id": "3173146", "postDate": "04/07/2025 15:24:21", "content": "It's been ~3 hours since the last update, and there's still not a single participant on the leaderboard. perhaps the leaderboard has lost 'weight' not only for you and me)))
\nStill, it's a lesson for all of us to duplicate datasets locally and always use try… except block in notebooks
", "rawMarkdown": "It's been ~3 hours since the last update, and there's still not a single participant on the leaderboard. perhaps the leaderboard has lost 'weight' not only for you and me)))\n\nStill, it's a lesson for all of us to duplicate datasets locally and always use try... except block in notebooks", "votes": null }, { "id": "3173156", "postDate": "04/07/2025 15:36:05", "content": "I took this error into account in AIMO-2 and in the end, those 2 notebooks that I was going to submit to LB, I replaced the public models with my duplicates. And I put a try…except block. If a critical error occurs on some answer, then I will only lose one answer, but the notebook will continue to run.
", "rawMarkdown": "I took this error into account in AIMO-2 and in the end, those 2 notebooks that I was going to submit to LB, I replaced the public models with my duplicates. And I put a try...except block. If a critical error occurs on some answer, then I will only lose one answer, but the notebook will continue to run.", "votes": null }, { "id": "3173190", "postDate": "04/07/2025 16:17:57", "content": "I don't think it's your fault. Kaggle is supposed to take a snapshot of your submission so that neither you nor anyone else (in case you're using public notebooks/models) can influence the result. I would recommend reaching out to the staff—maybe they have a way to restore the model.
", "rawMarkdown": "I don't think it's your fault. Kaggle is supposed to take a snapshot of your submission so that neither you nor anyone else (in case you're using public notebooks/models) can influence the result. I would recommend reaching out to the staff—maybe they have a way to restore the model.", "votes": null }, { "id": "3173209", "postDate": "04/07/2025 16:36:53", "content": "", "rawMarkdown": "Wrote.\nhttps://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/572070#3173200", "votes": null }, { "id": "3175035", "postDate": "04/09/2025 16:56:06", "content": "Kaggle Support (Kaggle)
\nApr 9, 2025, 9:45 AM PDT
\nHello Konstantin,
Thank you for your inquiry. Once the data is deleted from Kaggle we cannot recover the data.
\nYour request (177296) has been updated. To add additional comments, reply to this email.
I also made a request in Feedback to add data deletion protection properties. Let's see what comes of it.
\nhttps://www.kaggle.com/discussions/product-feedback/572396
", "rawMarkdown": "I also made a request in Feedback to add data deletion protection properties. Let's see what comes of it.\n\nhttps://www.kaggle.com/discussions/product-feedback/572396", "votes": null }, { "id": "3175488", "postDate": "04/10/2025 07:40:49", "content": "There is some positive news, I was heard on the Feedback page. I am starting the process of restoring the models. Let's try to restore the notebooks. And in the future they promised to make a basket, like Windows, with a validity period of one month.
", "rawMarkdown": "There is some positive news, I was heard on the Feedback page. I am starting the process of restoring the models. Let's try to restore the notebooks. And in the future they promised to make a basket, like Windows, with a validity period of one month.", "votes": null }, { "id": "3177098", "postDate": "04/12/2025 09:02:48", "content": "https://www.kaggle.com/discussions/product-feedback/572396#3176911
", "rawMarkdown": "https://www.kaggle.com/discussions/product-feedback/572396#3176911", "votes": null } ], "comments": [ { "id": 3173047, "author_name": "hertryggr", "author_url": "", "post_date": "04/07/2025 12:51:50", "content": "Hi, Konstantin. Everything happens for the first time, and you shouldn't blame yourself for mistakes (´-ω-`)
\nBut I'm burning with curiosity about how much \"weight' the leaderboard will lose))) (i also got an error, i used these deleted datasets)
", "votes": null, "replies": [ { "id": 3173122, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/07/2025 14:42:58", "content": "Hi @hertryggr. In theory, it has lost weight now, I am no longer on the list, and therefore, perhaps, someone else is, if my models were used in both versions and there was no try … except block.
\nI was so busy with the AIMO-2 competition, I sat until three in the morning and I didn’t have enough space for models and datasets, and there were models from 20 to 60 GB, constant quantization of models, I eventually filled both data warehouses. The most interesting thing is that I thought that the competition was already over and I didn’t even get bronze. Sometimes such glitches happen in the head. Well, accordingly, I smashed all the models. And then I saw on the forum that you can view a hidden submission bookmark. And when I looked, I was shocked. There was absolutely no working notebook that was without errors. I tried a bunch of models from github to get something good. I created 225 versions of notebooks in experiments. I really spent 2 months on this. And I also deleted all these versions from the local computer, since LLM models filled up 2 TB of SSD. (The non-quantized LLM model takes up to 160 GB of space.) In other words, I don't have any models from this contest on the site or on my local computer. Only the notebooks remain.
\nNow I understand that I was right, I definitely won’t even get bronze now.
", "votes": null, "replies": [ { "id": 3173146, "author_name": "hertryggr", "author_url": "", "post_date": "04/07/2025 15:24:21", "content": "It's been ~3 hours since the last update, and there's still not a single participant on the leaderboard. perhaps the leaderboard has lost 'weight' not only for you and me)))
\nStill, it's a lesson for all of us to duplicate datasets locally and always use try… except block in notebooks
", "votes": null, "replies": [ { "id": 3173156, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/07/2025 15:36:05", "content": "I took this error into account in AIMO-2 and in the end, those 2 notebooks that I was going to submit to LB, I replaced the public models with my duplicates. And I put a try…except block. If a critical error occurs on some answer, then I will only lose one answer, but the notebook will continue to run.
", "votes": null, "replies": [] } ] } ] } ] }, { "id": 3173190, "author_name": "skrrydg", "author_url": "", "post_date": "04/07/2025 16:17:57", "content": "I don't think it's your fault. Kaggle is supposed to take a snapshot of your submission so that neither you nor anyone else (in case you're using public notebooks/models) can influence the result. I would recommend reaching out to the staff—maybe they have a way to restore the model.
", "votes": null, "replies": [ { "id": 3173209, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/07/2025 16:36:53", "content": "", "votes": null, "replies": [] } ] }, { "id": 3175035, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/09/2025 16:56:06", "content": "Kaggle Support (Kaggle)
\nApr 9, 2025, 9:45 AM PDT
\nHello Konstantin,
Thank you for your inquiry. Once the data is deleted from Kaggle we cannot recover the data.
\nYour request (177296) has been updated. To add additional comments, reply to this email.
I also made a request in Feedback to add data deletion protection properties. Let's see what comes of it.
\nhttps://www.kaggle.com/discussions/product-feedback/572396
", "votes": null, "replies": [ { "id": 3175488, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/10/2025 07:40:49", "content": "There is some positive news, I was heard on the Feedback page. I am starting the process of restoring the models. Let's try to restore the notebooks. And in the future they promised to make a basket, like Windows, with a validity period of one month.
", "votes": null, "replies": [ { "id": 3177098, "author_name": "konstantinboyko", "author_url": "", "post_date": "04/12/2025 09:02:48", "content": "https://www.kaggle.com/discussions/product-feedback/572396#3176911
", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3172329": "I accidentally deleted my data models to make room for new tasks, forgetting that the competition was not over yet. As a result, all my notebooks turned into pumpkins. And I was eliminated from the competition for this reason. I am very upset and sad, but it is my fault.\n\nI apologize to those who used my datasets from a public notebook. This happened to me for the first time. Next time, I will not repeat this. I will remember my mistake for a long time.\n\n\n\n", "3173047": "Hi, Konstantin. Everything happens for the first time, and you shouldn't blame yourself for mistakes (´-ω-`)\n\nBut I'm burning with curiosity about how much \"weight' the leaderboard will lose))) (i also got an error, i used these deleted datasets)", "3173122": "Hi @hertryggr. In theory, it has lost weight now, I am no longer on the list, and therefore, perhaps, someone else is, if my models were used in both versions and there was no try ... except block. \n\nI was so busy with the AIMO-2 competition, I sat until three in the morning and I didn’t have enough space for models and datasets, and there were models from 20 to 60 GB, constant quantization of models, I eventually filled both data warehouses. The most interesting thing is that I thought that the competition was already over and I didn’t even get bronze. Sometimes such glitches happen in the head. Well, accordingly, I smashed all the models. And then I saw on the forum that you can view a hidden submission bookmark. And when I looked, I was shocked. There was absolutely no working notebook that was without errors. I tried a bunch of models from github to get something good. I created 225 versions of notebooks in experiments. I really spent 2 months on this. And I also deleted all these versions from the local computer, since LLM models filled up 2 TB of SSD. (The non-quantized LLM model takes up to 160 GB of space.) In other words, I don't have any models from this contest on the site or on my local computer. Only the notebooks remain.\n\nNow I understand that I was right, I definitely won’t even get bronze now.", "3173146": "It's been ~3 hours since the last update, and there's still not a single participant on the leaderboard. perhaps the leaderboard has lost 'weight' not only for you and me)))\n\nStill, it's a lesson for all of us to duplicate datasets locally and always use try... except block in notebooks", "3173156": "I took this error into account in AIMO-2 and in the end, those 2 notebooks that I was going to submit to LB, I replaced the public models with my duplicates. And I put a try...except block. If a critical error occurs on some answer, then I will only lose one answer, but the notebook will continue to run.", "3173190": "I don't think it's your fault. Kaggle is supposed to take a snapshot of your submission so that neither you nor anyone else (in case you're using public notebooks/models) can influence the result. I would recommend reaching out to the staff—maybe they have a way to restore the model.", "3173209": "Wrote.\nhttps://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/572070#3173200", "3175035": "Kaggle Support (Kaggle)\n\nApr 9, 2025, 9:45 AM PDT\nHello Konstantin,\n \nThank you for your inquiry. Once the data is deleted from Kaggle we cannot recover the data. \n\nYour request (177296) has been updated. To add additional comments, reply to this email.", "3175038": "I also made a request in Feedback to add data deletion protection properties. Let's see what comes of it.\n\nhttps://www.kaggle.com/discussions/product-feedback/572396", "3175488": "There is some positive news, I was heard on the Feedback page. I am starting the process of restoring the models. Let's try to restore the notebooks. And in the future they promised to make a basket, like Windows, with a validity period of one month.", "3177098": "https://www.kaggle.com/discussions/product-feedback/572396#3176911" }, "source": "meta" }