{ "id": 77774, "title": "What's your local CV score at 0.7 LB ?", "url": "/competitions/quora-insincere-questions-classification/discussion/77774", "author_name": "", "post_date": "2019-01-16T16:00:11.523286500Z", "votes": 17, "comment_count": 19, "views": 0, "content": "

Can the people on 0.700 on the public LB kindly share their local CV scores. I'm at 0.6811 (5 fold). Whenever my local CV increases above 0.6811, my public LB falls to 0.695 or something. Should I trust my local CV over the public LB ?

", "messages": [ { "id": "456822", "postDate": "01/16/2019 16:00:11", "content": "

Can the people on 0.700 on the public LB kindly share their local CV scores. I'm at 0.6811 (5 fold). Whenever my local CV increases above 0.6811, my public LB falls to 0.695 or something. Should I trust my local CV over the public LB ?

", "rawMarkdown": "Can the people on 0.700 on the public LB kindly share their local CV scores. I'm at 0.6811 (5 fold). Whenever my local CV increases above 0.6811, my public LB falls to 0.695 or something. Should I trust my local CV over the public LB ?", "votes": null }, { "id": "457267", "postDate": "01/17/2019 07:08:56", "content": "

my score \nCV :0.6803\nLB :0.7\nI got the same observation with you about the 0.6811 thing. I dont know whether should I trust the my CV or LB now haha

", "rawMarkdown": "my score \nCV :0.6803\nLB :0.7\nI got the same observation with you about the 0.6811 thing. I dont know whether should I trust the my CV or LB now haha", "votes": null }, { "id": "457283", "postDate": "01/17/2019 07:26:17", "content": "

0.6749

", "rawMarkdown": "0.6749", "votes": null }, { "id": "457305", "postDate": "01/17/2019 07:52:05", "content": "

CV: 0.6725 (5 fold), LB:0.700\nIt's an overfitting result though.

", "rawMarkdown": "CV: 0.6725 (5 fold), LB:0.700\nIt's an overfitting result though.", "votes": null }, { "id": "457521", "postDate": "01/17/2019 15:51:56", "content": "

cv 0.6884\nlb 0.7\nb.t.w relation between my cv and lb is very strange,i don't have any idea about this

", "rawMarkdown": "cv 0.6884\nlb 0.7\nb.t.w relation between my cv and lb is very strange,i don't have any idea about this", "votes": null }, { "id": "457536", "postDate": "01/17/2019 16:44:04", "content": "

Admittedly I've only made a single submission as I'm just starting but am getting a CV over 0.69 so far.

", "rawMarkdown": "Admittedly I've only made a single submission as I'm just starting but am getting a CV over 0.69 so far.", "votes": null }, { "id": "457707", "postDate": "01/18/2019 00:53:53", "content": "

cv 693 lb 697\ncv 691 lb 701\ncv 692 lb 702\ncv 690 lb701\ncv 690 lb 697\ncv 695 lb 693\nUPADATE:\ncv 698 lb still 695\ncv 699 lb still 694....\ncv 700 lb still 694....

", "rawMarkdown": "cv 693 lb 697\ncv 691 lb 701\ncv 692 lb 702\ncv 690 lb701\ncv 690 lb 697\ncv 695 lb 693\nUPADATE:\ncv 698 lb still 695\ncv 699 lb still 694....\ncv 700 lb still 694....", "votes": null }, { "id": "457745", "postDate": "01/18/2019 02:55:44", "content": "

my score \nCV :0.6795\nLB :0.7

", "rawMarkdown": "my score \nCV :0.6795\nLB :0.7", "votes": null }, { "id": "457769", "postDate": "01/18/2019 04:24:17", "content": "

my score: CV --> 0.6782; LB --> 0.703

", "rawMarkdown": "my score: CV --> 0.6782; LB --> 0.703", "votes": null }, { "id": "457822", "postDate": "01/18/2019 06:12:55", "content": "

Hi, I notice that val score is very similar to your public kernel. May I ask what changed to get you up on LB? It doesn't have to be very specific, just the direction. Maybe more preprocessing or a different architecture or ensembling?

", "rawMarkdown": "Hi, I notice that val score is very similar to your public kernel. May I ask what changed to get you up on LB? It doesn't have to be very specific, just the direction. Maybe more preprocessing or a different architecture or ensembling?", "votes": null }, { "id": "457971", "postDate": "01/18/2019 12:53:56", "content": "

I got local 0.692-> LB0.702

", "rawMarkdown": "I got local 0.692-> LB0.702", "votes": null }, { "id": "459653", "postDate": "01/22/2019 06:11:52", "content": "

CV 0.6777 LB 0.704\namazing,trying to reproduce it

", "rawMarkdown": "CV 0.6777 LB 0.704\namazing,trying to reproduce it", "votes": null }, { "id": "459898", "postDate": "01/22/2019 15:45:13", "content": "

CV 0.6786 LB 0.701

", "rawMarkdown": "CV 0.6786 LB 0.701", "votes": null }, { "id": "459942", "postDate": "01/22/2019 16:55:28", "content": "

Really strong cv

", "rawMarkdown": "Really strong cv", "votes": null }, { "id": "459946", "postDate": "01/22/2019 17:01:22", "content": "

sorry, it was 0.6786, my mistake.

", "rawMarkdown": "sorry, it was 0.6786, my mistake.", "votes": null }, { "id": "459965", "postDate": "01/22/2019 17:56:20", "content": "

CV 0.6811, LB 0.702

", "rawMarkdown": "CV 0.6811, LB 0.702", "votes": null }, { "id": "459970", "postDate": "01/22/2019 18:14:14", "content": "

no problem makes sense now.

", "rawMarkdown": "no problem makes sense now.", "votes": null }, { "id": "464033", "postDate": "01/31/2019 05:17:00", "content": "

CV 0.6909 loss 0.0958 correlation 0.9515, LB 0.702

\n\n

Another submission CV 0.6804, LB 0.702.

\n\n

The first one is only slightly better than the second. :(

", "rawMarkdown": "CV 0.6909 loss 0.0958 correlation 0.9515, LB 0.702\n\nAnother submission CV 0.6804, LB 0.702.\n\nThe first one is only slightly better than the second. :(", "votes": null }, { "id": "464110", "postDate": "01/31/2019 08:14:01", "content": "

cv = 0.696892\nlb = 0.702

", "rawMarkdown": "cv = 0.696892\nlb = 0.702", "votes": null }, { "id": "465445", "postDate": "02/03/2019 05:48:56", "content": "

I have two models\ncv 683 lb 704\ncv 679 lb 705

", "rawMarkdown": "I have two models\ncv 683 lb 704\ncv 679 lb 705", "votes": null } ], "comments": [ { "id": 457267, "author_name": "dttung2905", "author_url": "", "post_date": "01/17/2019 07:08:56", "content": "

my score \nCV :0.6803\nLB :0.7\nI got the same observation with you about the 0.6811 thing. I dont know whether should I trust the my CV or LB now haha

", "votes": null, "replies": [] }, { "id": 457283, "author_name": "shujian", "author_url": "", "post_date": "01/17/2019 07:26:17", "content": "

0.6749

", "votes": null, "replies": [ { "id": 457822, "author_name": "wenrui29", "author_url": "", "post_date": "01/18/2019 06:12:55", "content": "

Hi, I notice that val score is very similar to your public kernel. May I ask what changed to get you up on LB? It doesn't have to be very specific, just the direction. Maybe more preprocessing or a different architecture or ensembling?

", "votes": null, "replies": [] } ] }, { "id": 457305, "author_name": "playif1", "author_url": "", "post_date": "01/17/2019 07:52:05", "content": "

CV: 0.6725 (5 fold), LB:0.700\nIt's an overfitting result though.

", "votes": null, "replies": [] }, { "id": 457521, "author_name": "zengyao", "author_url": "", "post_date": "01/17/2019 15:51:56", "content": "

cv 0.6884\nlb 0.7\nb.t.w relation between my cv and lb is very strange,i don't have any idea about this

", "votes": null, "replies": [] }, { "id": 457536, "author_name": "maw501", "author_url": "", "post_date": "01/17/2019 16:44:04", "content": "

Admittedly I've only made a single submission as I'm just starting but am getting a CV over 0.69 so far.

", "votes": null, "replies": [] }, { "id": 457707, "author_name": "baomengjiao", "author_url": "", "post_date": "01/18/2019 00:53:53", "content": "

cv 693 lb 697\ncv 691 lb 701\ncv 692 lb 702\ncv 690 lb701\ncv 690 lb 697\ncv 695 lb 693\nUPADATE:\ncv 698 lb still 695\ncv 699 lb still 694....\ncv 700 lb still 694....

", "votes": null, "replies": [] }, { "id": 457745, "author_name": "alexxanderlarko", "author_url": "", "post_date": "01/18/2019 02:55:44", "content": "

my score \nCV :0.6795\nLB :0.7

", "votes": null, "replies": [] }, { "id": 457769, "author_name": "", "author_url": "", "post_date": "01/18/2019 04:24:17", "content": "

my score: CV --> 0.6782; LB --> 0.703

", "votes": null, "replies": [] }, { "id": 457971, "author_name": "vanche", "author_url": "", "post_date": "01/18/2019 12:53:56", "content": "

I got local 0.692-> LB0.702

", "votes": null, "replies": [] }, { "id": 459653, "author_name": "peining", "author_url": "", "post_date": "01/22/2019 06:11:52", "content": "

CV 0.6777 LB 0.704\namazing,trying to reproduce it

", "votes": null, "replies": [] }, { "id": 459898, "author_name": "garybios", "author_url": "", "post_date": "01/22/2019 15:45:13", "content": "

CV 0.6786 LB 0.701

", "votes": null, "replies": [ { "id": 459942, "author_name": "satian", "author_url": "", "post_date": "01/22/2019 16:55:28", "content": "

Really strong cv

", "votes": null, "replies": [] }, { "id": 459946, "author_name": "garybios", "author_url": "", "post_date": "01/22/2019 17:01:22", "content": "

sorry, it was 0.6786, my mistake.

", "votes": null, "replies": [] }, { "id": 459970, "author_name": "satian", "author_url": "", "post_date": "01/22/2019 18:14:14", "content": "

no problem makes sense now.

", "votes": null, "replies": [] } ] }, { "id": 459965, "author_name": "matthewygf", "author_url": "", "post_date": "01/22/2019 17:56:20", "content": "

CV 0.6811, LB 0.702

", "votes": null, "replies": [] }, { "id": 464033, "author_name": "zhutianyucs", "author_url": "", "post_date": "01/31/2019 05:17:00", "content": "

CV 0.6909 loss 0.0958 correlation 0.9515, LB 0.702

\n\n

Another submission CV 0.6804, LB 0.702.

\n\n

The first one is only slightly better than the second. :(

", "votes": null, "replies": [] }, { "id": 464110, "author_name": "springmanndaniel", "author_url": "", "post_date": "01/31/2019 08:14:01", "content": "

cv = 0.696892\nlb = 0.702

", "votes": null, "replies": [] }, { "id": 465445, "author_name": "jetouxu", "author_url": "", "post_date": "02/03/2019 05:48:56", "content": "

I have two models\ncv 683 lb 704\ncv 679 lb 705

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "456822": "Can the people on 0.700 on the public LB kindly share their local CV scores. I'm at 0.6811 (5 fold). Whenever my local CV increases above 0.6811, my public LB falls to 0.695 or something. Should I trust my local CV over the public LB ?", "457267": "my score \nCV :0.6803\nLB :0.7\nI got the same observation with you about the 0.6811 thing. I dont know whether should I trust the my CV or LB now haha", "457283": "0.6749", "457305": "CV: 0.6725 (5 fold), LB:0.700\nIt's an overfitting result though.", "457521": "cv 0.6884\nlb 0.7\nb.t.w relation between my cv and lb is very strange,i don't have any idea about this", "457536": "Admittedly I've only made a single submission as I'm just starting but am getting a CV over 0.69 so far.", "457707": "cv 693 lb 697\ncv 691 lb 701\ncv 692 lb 702\ncv 690 lb701\ncv 690 lb 697\ncv 695 lb 693\nUPADATE:\ncv 698 lb still 695\ncv 699 lb still 694....\ncv 700 lb still 694....", "457745": "my score \nCV :0.6795\nLB :0.7", "457769": "my score: CV --> 0.6782; LB --> 0.703", "457822": "Hi, I notice that val score is very similar to your public kernel. May I ask what changed to get you up on LB? It doesn't have to be very specific, just the direction. Maybe more preprocessing or a different architecture or ensembling?", "457971": "I got local 0.692-> LB0.702", "459653": "CV 0.6777 LB 0.704\namazing,trying to reproduce it", "459898": "CV 0.6786 LB 0.701", "459942": "Really strong cv", "459946": "sorry, it was 0.6786, my mistake.", "459965": "CV 0.6811, LB 0.702", "459970": "no problem makes sense now.", "464033": "CV 0.6909 loss 0.0958 correlation 0.9515, LB 0.702\n\nAnother submission CV 0.6804, LB 0.702.\n\nThe first one is only slightly better than the second. :(", "464110": "cv = 0.696892\nlb = 0.702", "465445": "I have two models\ncv 683 lb 704\ncv 679 lb 705" }, "source": "meta" }