First, congratulations wherever you landed if you had one of the top PyTorch scores on Private LB! I am looking forward to learning from anyone who shares their notebook. I would really like to see so I can do a GAP analysis and understand where I should have spent more time, what I should have done differently. Regardless, I learned so much, and I am a huge fan of PyTorch and the PyTorch community.
\nAgain, sincere congratulations to anyone who medaled with PyTorch!
", "messages": [ { "id": "974447", "postDate": "08/18/2020 00:27:04", "content": "First, congratulations wherever you landed if you had one of the top PyTorch scores on Private LB! I am looking forward to learning from anyone who shares their notebook. I would really like to see so I can do a GAP analysis and understand where I should have spent more time, what I should have done differently. Regardless, I learned so much, and I am a huge fan of PyTorch and the PyTorch community.
\nAgain, sincere congratulations to anyone who medaled with PyTorch!
", "rawMarkdown": "First, congratulations wherever you landed if you had one of the top PyTorch scores on Private LB! I am looking forward to learning from anyone who shares their notebook. I would really like to see so I can do a GAP analysis and understand where I should have spent more time, what I should have done differently. Regardless, I learned so much, and I am a huge fan of PyTorch and the PyTorch community. \n\nAgain, sincere congratulations to anyone who medaled with PyTorch!", "votes": null }, { "id": "974452", "postDate": "08/18/2020 00:28:02", "content": "I used PyTorch for this competition. My best single model scored 0.9493 on public LB and 0.9495 on private LB (but I selected the ensemble instead, which was worse)! EfficientNet-B6, trained on pseudolabeled data.
", "rawMarkdown": "I used PyTorch for this competition. My best single model scored 0.9493 on public LB and 0.9495 on private LB (but I selected the ensemble instead, which was worse)! EfficientNet-B6, trained on pseudolabeled data.", "votes": null }, { "id": "974459", "postDate": "08/18/2020 00:29:27", "content": "I got 95.6 public LB and 93.31 private LB with a single PyTorch model. I'll be doing a detailed writeup + releasing all of my code very soon! :)
", "rawMarkdown": "I got 95.6 public LB and 93.31 private LB with a single PyTorch model. I'll be doing a detailed writeup + releasing all of my code very soon! :)", "votes": null }, { "id": "974470", "postDate": "08/18/2020 00:38:11", "content": "@vaillant would be interesting to see how you came up with your Psuedolabels. That's an amazingly tight CV btw, were all your Public/Private CV's that close to each other?
", "rawMarkdown": "vaillant would be interesting to see how you came up with your Psuedolabels. That's an amazingly tight CV btw, were all your Public/Private CV's that close to each other?", "votes": null }, { "id": "975151", "postDate": "08/18/2020 07:12:45", "content": "great! I can see my learning experiences will start soon. :)
", "rawMarkdown": "great! I can see my learning experiences will start soon. :)", "votes": null } ], "comments": [ { "id": 974452, "author_name": "vaillant", "author_url": "", "post_date": "08/18/2020 00:28:02", "content": "I used PyTorch for this competition. My best single model scored 0.9493 on public LB and 0.9495 on private LB (but I selected the ensemble instead, which was worse)! EfficientNet-B6, trained on pseudolabeled data.
", "votes": null, "replies": [ { "id": 974470, "author_name": "brianfeeny", "author_url": "", "post_date": "08/18/2020 00:38:11", "content": "@vaillant would be interesting to see how you came up with your Psuedolabels. That's an amazingly tight CV btw, were all your Public/Private CV's that close to each other?
", "votes": null, "replies": [] } ] }, { "id": 974459, "author_name": "themlenthusiast", "author_url": "", "post_date": "08/18/2020 00:29:27", "content": "I got 95.6 public LB and 93.31 private LB with a single PyTorch model. I'll be doing a detailed writeup + releasing all of my code very soon! :)
", "votes": null, "replies": [] }, { "id": 975151, "author_name": "yimacs", "author_url": "", "post_date": "08/18/2020 07:12:45", "content": "great! I can see my learning experiences will start soon. :)
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "974447": "First, congratulations wherever you landed if you had one of the top PyTorch scores on Private LB! I am looking forward to learning from anyone who shares their notebook. I would really like to see so I can do a GAP analysis and understand where I should have spent more time, what I should have done differently. Regardless, I learned so much, and I am a huge fan of PyTorch and the PyTorch community. \n\nAgain, sincere congratulations to anyone who medaled with PyTorch!", "974452": "I used PyTorch for this competition. My best single model scored 0.9493 on public LB and 0.9495 on private LB (but I selected the ensemble instead, which was worse)! EfficientNet-B6, trained on pseudolabeled data.", "974459": "I got 95.6 public LB and 93.31 private LB with a single PyTorch model. I'll be doing a detailed writeup + releasing all of my code very soon! :)", "974470": "vaillant would be interesting to see how you came up with your Psuedolabels. That's an amazingly tight CV btw, were all your Public/Private CV's that close to each other?", "975151": "great! I can see my learning experiences will start soon. :)" }, "source": "meta" }