I'm new to competitions, so apologies if this is a silly question. Can we use a separate notebook for preprocessing/training the model, save the model, and use it in a notebook dedicated for submission? Or does our preprocessing/training have to be included in our submission?
\nI ask because it effects the run-time of the notebook.
\nThanks in advance!
I'm new to competitions, so apologies if this is a silly question. Can we use a separate notebook for preprocessing/training the model, save the model, and use it in a notebook dedicated for submission? Or does our preprocessing/training have to be included in our submission?
\nI ask because it effects the run-time of the notebook.
\nThanks in advance!
You can save output of notebooks and then use them as dataset in another notebook using '+ Add Data' tab on the right side of the notebook. This way you can create chains of notebook from data creation to training to inference to submission. Very handy.
", "rawMarkdown": "You can save output of notebooks and then use them as dataset in another notebook using '+ Add Data' tab on the right side of the notebook. This way you can create chains of notebook from data creation to training to inference to submission. Very handy." }, { "id": 1926648, "postDate": "2022-09-05T02:40:31.770Z", "content": "Of course you could. What I did is to train the model locally, and then upload the weights as a dataset. In the submission version of notebook, the only thing I need to do is load that weights from dataset and do the inference
", "rawMarkdown": "Of course you could. What I did is to train the model locally, and then upload the weights as a dataset. In the submission version of notebook, the only thing I need to do is load that weights from dataset and do the inference" }, { "id": 1923257, "postDate": "2022-09-02T05:06:33.057Z", "content": "Thanks for asking this question. As a newbie, I was wondering the same.
", "rawMarkdown": "Thanks for asking this question. As a newbie, I was wondering the same. " }, { "id": 1922531, "postDate": "2022-09-01T14:51:49.383Z", "content": "Yes, It is done all the time
", "rawMarkdown": "Yes, It is done all the time " }, { "id": 1917115, "postDate": "2022-08-28T12:51:44.040Z", "content": "Hi @robinonsay1,
\nYes, a separate notebook dedicated just for inference(with pre-trained weights) can be used for submission.
", "rawMarkdown": "Hi @robinonsay1,\n\nYes, a separate notebook dedicated just for inference(with pre-trained weights) can be used for submission." } ], "comments": [ { "id": 1946173, "author_name": "tdiceman", "author_url": "", "post_date": "2022-09-19T16:32:09.737000", "content": "You can save output of notebooks and then use them as dataset in another notebook using '+ Add Data' tab on the right side of the notebook. This way you can create chains of notebook from data creation to training to inference to submission. Very handy.
", "votes": 0, "replies": [] }, { "id": 1926648, "author_name": "Chen Zhang", "author_url": "", "post_date": "2022-09-05T02:40:31.770000", "content": "Of course you could. What I did is to train the model locally, and then upload the weights as a dataset. In the submission version of notebook, the only thing I need to do is load that weights from dataset and do the inference
", "votes": 0, "replies": [] }, { "id": 1923257, "author_name": "Saurabh Sawhney", "author_url": "", "post_date": "2022-09-02T05:06:33.057000", "content": "Thanks for asking this question. As a newbie, I was wondering the same.
", "votes": 0, "replies": [] }, { "id": 1922531, "author_name": "The Devastator", "author_url": "", "post_date": "2022-09-01T14:51:49.383000", "content": "Yes, It is done all the time
", "votes": 0, "replies": [] }, { "id": 1917115, "author_name": "Yashika Jain", "author_url": "", "post_date": "2022-08-28T12:51:44.040000", "content": "Hi @robinonsay1,
\nYes, a separate notebook dedicated just for inference(with pre-trained weights) can be used for submission.
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1916488": "I'm new to competitions, so apologies if this is a silly question. Can we use a separate notebook for preprocessing/training the model, save the model, and use it in a notebook dedicated for submission? Or does our preprocessing/training have to be included in our submission?\n\nI ask because it effects the run-time of the notebook.\nThanks in advance!", "1946173": "You can save output of notebooks and then use them as dataset in another notebook using '+ Add Data' tab on the right side of the notebook. This way you can create chains of notebook from data creation to training to inference to submission. Very handy.", "1926648": "Of course you could. What I did is to train the model locally, and then upload the weights as a dataset. In the submission version of notebook, the only thing I need to do is load that weights from dataset and do the inference", "1923257": "Thanks for asking this question. As a newbie, I was wondering the same. ", "1922531": "Yes, It is done all the time ", "1917115": "Hi @robinonsay1,\n\nYes, a separate notebook dedicated just for inference(with pre-trained weights) can be used for submission." } }