Did anyone encount this error???\nI want to share about Notebook Exceeded Allowed Compute Error.
\n\nI published my problematic notebook is following.
\nhttps://www.kaggle.com/nayuts/landmark-resnet50-wip
@nayuts I might be mistaken, but I think you might run out of memory. I see that your output is of size 81313. This output will be used as a feature embedding/descriptor for each image, and compared with other embeddings of other images. I'm using a triplet loss as the learning objective.
\n\nEDIT (side note): I'm new to this type of machine learning problem. The first learning objective (for this type of problem) that I stumbled upon happen to be the triplet loss. However, the triplet loss might not be the best choice here (?). There seems to be many variants out there (I guess we'll need to dig through a lot of literature)
", "rawMarkdown": "@nayuts I might be mistaken, but I think you might run out of memory. I see that your output is of size 81313. This output will be used as a feature embedding/descriptor for each image, and compared with other embeddings of other images. I'm using a [triplet loss](https://www.youtube.com/watch?v=d2XB5-tuCWU) as the learning objective.\n\nEDIT (side note): I'm new to this type of machine learning problem. The first learning objective (for this type of problem) that I stumbled upon happen to be the triplet loss. However, the triplet loss might not be the best choice here (?). There seems to be many variants out there (I guess we'll need to dig through a lot of literature)", "votes": 1, "replies": [ { "id": 922604, "postDate": "2020-07-10T08:23:37.013Z", "rawMarkdown": "", "isDeleted": true }, { "id": 922823, "postDate": "2020-07-10T10:57:19.637Z", "content": "Thank you for your great advices! \nImmediately, I watched the youtube content you told me!\nI was not familiar with solutions which adopt loss functions such as triplet loss.\nSo I was wondering why someone can adopt triplet loss soon!\nI'll use the loss for my notebook! So thank you!!!
", "rawMarkdown": "Thank you for your great advices! \nImmediately, I watched the youtube content you told me!\nI was not familiar with solutions which adopt loss functions such as triplet loss.\nSo I was wondering why someone can adopt triplet loss soon!\nI'll use the loss for my notebook! So thank you!!!\n" } ] }, { "id": 921774, "postDate": "2020-07-09T14:42:25.967Z", "content": "Did anyone encount this error???\nI want to share about Notebook Exceeded Allowed Compute Error.
\n\nI published my problematic notebook is following.
\nhttps://www.kaggle.com/nayuts/landmark-resnet50-wip
@nayuts I might be mistaken, but I think you might run out of memory. I see that your output is of size 81313. This output will be used as a feature embedding/descriptor for each image, and compared with other embeddings of other images. I'm using a triplet loss as the learning objective.
\n\nEDIT (side note): I'm new to this type of machine learning problem. The first learning objective (for this type of problem) that I stumbled upon happen to be the triplet loss. However, the triplet loss might not be the best choice here (?). There seems to be many variants out there (I guess we'll need to dig through a lot of literature)
", "votes": 1, "replies": [ { "id": 922604, "author_name": "", "author_url": "", "post_date": "2020-07-10T08:23:37.013000", "content": "", "votes": 0, "replies": [] }, { "id": 922823, "author_name": "Nayu.T.S.", "author_url": "", "post_date": "2020-07-10T10:57:19.637000", "content": "Thank you for your great advices! \nImmediately, I watched the youtube content you told me!\nI was not familiar with solutions which adopt loss functions such as triplet loss.\nSo I was wondering why someone can adopt triplet loss soon!\nI'll use the loss for my notebook! So thank you!!!
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "922206": "@nayuts I might be mistaken, but I think you might run out of memory. I see that your output is of size 81313. This output will be used as a feature embedding/descriptor for each image, and compared with other embeddings of other images. I'm using a [triplet loss](https://www.youtube.com/watch?v=d2XB5-tuCWU) as the learning objective.\n\nEDIT (side note): I'm new to this type of machine learning problem. The first learning objective (for this type of problem) that I stumbled upon happen to be the triplet loss. However, the triplet loss might not be the best choice here (?). There seems to be many variants out there (I guess we'll need to dig through a lot of literature)", "921774": "Did anyone encount this error???\nI want to share about Notebook Exceeded Allowed Compute Error.\n\nI published my problematic notebook is following. \nhttps://www.kaggle.com/nayuts/landmark-resnet50-wip\n\n" } }