Hi to all!
\nDon't forget that for your notebook to be accepted in LB, it must be fully completed by the deadline. Therefore, in fact, the real deadline = the official deadline - execution time of your notebook.
\nI am writing about this because at one of the previous competitions not knowing this feature cost me about 100 places on LB :-)
\nGood luck to everyone and the right choice of Final Models: -) !!!
Hi to all!
\nDon't forget that for your notebook to be accepted in LB, it must be fully completed by the deadline. Therefore, in fact, the real deadline = the official deadline - execution time of your notebook.
\nI am writing about this because at one of the previous competitions not knowing this feature cost me about 100 places on LB :-)
\nGood luck to everyone and the right choice of Final Models: -) !!!
Hi Andrij, I am first time competing, what does that mean fully completed? Does that mean we need to put all together, the training notebook + the inference notebook in a single notebook, or it is enough to prepare only the inference notebook with the trained models inside as datasets? My inference notebook is running, 10 mins single model and the ensemble is running one hour?
\nCheers.
Hello! The point here is that when you click submit, your notebook starts up again and runs for a while. In order for it to be included in the overall standings, it must complete its work after clicking submit by the time of the official deadline.
", "rawMarkdown": "Hello! The point here is that when you click submit, your notebook starts up again and runs for a while. In order for it to be included in the overall standings, it must complete its work after clicking submit by the time of the official deadline.", "votes": null }, { "id": "1204409", "postDate": "02/16/2021 05:54:22", "content": "Hello, this is my offficial competetion. Do I need my kernel to reproduce my best submission? I have experimented with many different things and can't seem to reproduce the said submission. I am using different submission notebook and different training notebook. I am using the output of the training notebook as a dataset to predict, I have new versioned the dataset many times now.
", "rawMarkdown": "Hello, this is my offficial competetion. Do I need my kernel to reproduce my best submission? I have experimented with many different things and can't seem to reproduce the said submission. I am using different submission notebook and different training notebook. I am using the output of the training notebook as a dataset to predict, I have new versioned the dataset many times now.", "votes": null }, { "id": "1204840", "postDate": "02/16/2021 11:26:52", "content": "I'm a little bit confused. Isn't it OK to select 2 submissions with the highest LB scores as the \"Use for Final Score\"? Thanks.
", "rawMarkdown": "I'm a little bit confused. Isn't it OK to select 2 submissions with the highest LB scores as the \"Use for Final Score\"? Thanks.", "votes": null }, { "id": "1205015", "postDate": "02/16/2021 13:34:46", "content": "Hello! If you do not select any submissions kaggle automatically selects your best submissions based on the PUBLIC TEST SET. This does not mean that these will be the best submission on a private test іуе. As one grantmaster said, choosing the best submission is an art :-) Personally, I had both a drop of 1000 positions and a jump into the gold zone. Good luck!
", "rawMarkdown": "Hello! If you do not select any submissions kaggle automatically selects your best submissions based on the PUBLIC TEST SET. This does not mean that these will be the best submission on a private test іуе. As one grantmaster said, choosing the best submission is an art :-) Personally, I had both a drop of 1000 positions and a jump into the gold zone. Good luck!", "votes": null }, { "id": "1205021", "postDate": "02/16/2021 13:38:41", "content": "Hello! It all depends on your final place, if you win the competition as I understand it you will need to confirm your decision, but if you take the bronze I doubt that someone there will check your code. But of course it is better to keep notebooks with attempts, in the end it is unknown what will \"shoot\". Good luck!
", "rawMarkdown": "Hello! It all depends on your final place, if you win the competition as I understand it you will need to confirm your decision, but if you take the bronze I doubt that someone there will check your code. But of course it is better to keep notebooks with attempts, in the end it is unknown what will \"shoot\". Good luck!", "votes": null }, { "id": "1205358", "postDate": "02/16/2021 16:28:54", "content": "Oh ok. Well I am not going to win te competition, not even a bronze, so no problem then. I just learned a lot through this competition. I take it as a win.
", "rawMarkdown": "Oh ok. Well I am not going to win te competition, not even a bronze, so no problem then. I just learned a lot through this competition. I take it as a win.", "votes": null }, { "id": "1205513", "postDate": "02/16/2021 18:21:42", "content": "This is the essence of competition: to learn something new. Good luck!
", "rawMarkdown": "This is the essence of competition: to learn something new. Good luck!", "votes": null }, { "id": "1205740", "postDate": "02/17/2021 00:10:35", "content": "Thanks. Good luck!
", "rawMarkdown": "Thanks. Good luck!", "votes": null }, { "id": "1209617", "postDate": "02/19/2021 01:12:05", "content": "I am the guy who must response this topic.
\nThough I've seen your ddl notification the day before this day.
\nBut I still did't choose my best cv as my best two(I thought I could choose it today's morning , but I fell asleep). This guy is Private B 0.898(Medal zone).
\nAnd my another best score is Private B 0.898, if there is no other thing happens, this guy will be choosen automaticly as two best (::this guy has high Public scores ), but just exactly run out of the competition time (before ddl, mayebe just a few minutes).
\nThis makes me laugh from the morning until now. Lol.
\nFrom this game I've learned, nerve trust your Public socre, go head and trust discussion instead!!!!!!!!!!!!
Hi Andrij, I am first time competing, what does that mean fully completed? Does that mean we need to put all together, the training notebook + the inference notebook in a single notebook, or it is enough to prepare only the inference notebook with the trained models inside as datasets? My inference notebook is running, 10 mins single model and the ensemble is running one hour?
\nCheers.
Hello! The point here is that when you click submit, your notebook starts up again and runs for a while. In order for it to be included in the overall standings, it must complete its work after clicking submit by the time of the official deadline.
", "votes": null, "replies": [] } ] }, { "id": 1204409, "author_name": "mohneesh7", "author_url": "", "post_date": "02/16/2021 05:54:22", "content": "Hello, this is my offficial competetion. Do I need my kernel to reproduce my best submission? I have experimented with many different things and can't seem to reproduce the said submission. I am using different submission notebook and different training notebook. I am using the output of the training notebook as a dataset to predict, I have new versioned the dataset many times now.
", "votes": null, "replies": [ { "id": 1205021, "author_name": "aikhmelnytskyy", "author_url": "", "post_date": "02/16/2021 13:38:41", "content": "Hello! It all depends on your final place, if you win the competition as I understand it you will need to confirm your decision, but if you take the bronze I doubt that someone there will check your code. But of course it is better to keep notebooks with attempts, in the end it is unknown what will \"shoot\". Good luck!
", "votes": null, "replies": [] }, { "id": 1205358, "author_name": "mohneesh7", "author_url": "", "post_date": "02/16/2021 16:28:54", "content": "Oh ok. Well I am not going to win te competition, not even a bronze, so no problem then. I just learned a lot through this competition. I take it as a win.
", "votes": null, "replies": [] }, { "id": 1205513, "author_name": "aikhmelnytskyy", "author_url": "", "post_date": "02/16/2021 18:21:42", "content": "This is the essence of competition: to learn something new. Good luck!
", "votes": null, "replies": [] } ] }, { "id": 1204840, "author_name": "komakizzz", "author_url": "", "post_date": "02/16/2021 11:26:52", "content": "I'm a little bit confused. Isn't it OK to select 2 submissions with the highest LB scores as the \"Use for Final Score\"? Thanks.
", "votes": null, "replies": [ { "id": 1205015, "author_name": "aikhmelnytskyy", "author_url": "", "post_date": "02/16/2021 13:34:46", "content": "Hello! If you do not select any submissions kaggle automatically selects your best submissions based on the PUBLIC TEST SET. This does not mean that these will be the best submission on a private test іуе. As one grantmaster said, choosing the best submission is an art :-) Personally, I had both a drop of 1000 positions and a jump into the gold zone. Good luck!
", "votes": null, "replies": [] }, { "id": 1205740, "author_name": "komakizzz", "author_url": "", "post_date": "02/17/2021 00:10:35", "content": "Thanks. Good luck!
", "votes": null, "replies": [] } ] }, { "id": 1209617, "author_name": "fanwenping", "author_url": "", "post_date": "02/19/2021 01:12:05", "content": "I am the guy who must response this topic.
\nThough I've seen your ddl notification the day before this day.
\nBut I still did't choose my best cv as my best two(I thought I could choose it today's morning , but I fell asleep). This guy is Private B 0.898(Medal zone).
\nAnd my another best score is Private B 0.898, if there is no other thing happens, this guy will be choosen automaticly as two best (::this guy has high Public scores ), but just exactly run out of the competition time (before ddl, mayebe just a few minutes).
\nThis makes me laugh from the morning until now. Lol.
\nFrom this game I've learned, nerve trust your Public socre, go head and trust discussion instead!!!!!!!!!!!!