{
  "total": 130,
  "topics": [
    {
      "id": 40086,
      "title": "Hello from Cdiscount",
      "comment_count": 31,
      "views": 0,
      "votes": 24
    },
    {
      "id": 39465,
      "title": "External Data Thread",
      "comment_count": 56,
      "views": 0,
      "votes": 10
    },
    {
      "id": 39463,
      "title": "Welcome!",
      "comment_count": 22,
      "views": 0,
      "votes": 12
    },
    {
      "id": 45863,
      "title": "My brief summary of this competition[0.795]",
      "comment_count": 44,
      "views": 0,
      "votes": 134
    },
    {
      "id": 40498,
      "title": "pytorch starter kit here!",
      "comment_count": 63,
      "views": 0,
      "votes": 62
    },
    {
      "id": 41193,
      "title": "Loading Data blazing fast with MongoDB (<20ms~)",
      "comment_count": 15,
      "views": 0,
      "votes": 23
    },
    {
      "id": 41478,
      "title": "End-to-end image classification solution... and free $100 in gpu ",
      "comment_count": 64,
      "views": 0,
      "votes": 34
    },
    {
      "id": 39475,
      "title": "not install bson! install pymongo",
      "comment_count": 8,
      "views": 0,
      "votes": 12
    },
    {
      "id": 42454,
      "title": "Different ways to improve the class imbalance problem",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 45733,
      "title": "Our solution [5th place]",
      "comment_count": 17,
      "views": 0,
      "votes": 60
    },
    {
      "id": 53218,
      "title": "About the challenge data",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40780,
      "title": "batch size trick to accelerate training?",
      "comment_count": 26,
      "views": 0,
      "votes": 16
    },
    {
      "id": 41523,
      "title": "how to use hard samples?",
      "comment_count": 13,
      "views": 0,
      "votes": 4
    },
    {
      "id": 45721,
      "title": "My solution [9th place private]",
      "comment_count": 2,
      "views": 0,
      "votes": 56
    },
    {
      "id": 45737,
      "title": "Our 7th place solution",
      "comment_count": 3,
      "views": 0,
      "votes": 25
    },
    {
      "id": 39569,
      "title": "Hardware configurations",
      "comment_count": 11,
      "views": 0,
      "votes": 1
    },
    {
      "id": 50056,
      "title": "Video of 5th place solution with English subtitles",
      "comment_count": 0,
      "views": 0,
      "votes": 7
    },
    {
      "id": 45850,
      "title": "[0.776] Single model solution",
      "comment_count": 3,
      "views": 0,
      "votes": 29
    },
    {
      "id": 41021,
      "title": "Release of trained models (Cdiscount model zoo)",
      "comment_count": 52,
      "views": 0,
      "votes": 63
    },
    {
      "id": 40967,
      "title": "OSError: [Errno 28] No space left on device: ",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 45638,
      "title": "How do some top leaders submit their predictions so fast?",
      "comment_count": 10,
      "views": 0,
      "votes": 2
    },
    {
      "id": 45716,
      "title": "Congrats to new kaggle No.1",
      "comment_count": 19,
      "views": 0,
      "votes": 38
    },
    {
      "id": 45724,
      "title": "What hardware did you use?",
      "comment_count": 16,
      "views": 0,
      "votes": 16
    },
    {
      "id": 45756,
      "title": "Training in 8 epoch using focal loss",
      "comment_count": 2,
      "views": 0,
      "votes": 4
    },
    {
      "id": 45709,
      "title": "[ my ensemble solution  ]",
      "comment_count": 20,
      "views": 0,
      "votes": 49
    },
    {
      "id": 45774,
      "title": "GTX 1050 Ti solution [43rd place]",
      "comment_count": 0,
      "views": 0,
      "votes": 23
    },
    {
      "id": 45798,
      "title": "12th place solution",
      "comment_count": 9,
      "views": 0,
      "votes": 14
    },
    {
      "id": 45729,
      "title": "Single best model?",
      "comment_count": 10,
      "views": 0,
      "votes": 11
    },
    {
      "id": 45711,
      "title": "Thank You and Congratulations to All!",
      "comment_count": 4,
      "views": 0,
      "votes": 11
    },
    {
      "id": 45718,
      "title": "What I learnt in this competition and Congratulations to the winners.",
      "comment_count": 10,
      "views": 0,
      "votes": 4
    },
    {
      "id": 45717,
      "title": "Congratulations to the Winners!!!",
      "comment_count": 1,
      "views": 0,
      "votes": 15
    },
    {
      "id": 45710,
      "title": "Trainable and not trainable models",
      "comment_count": 4,
      "views": 0,
      "votes": 6
    },
    {
      "id": 45787,
      "title": "Transfer learning with NasNet and 2h Epochs",
      "comment_count": 0,
      "views": 0,
      "votes": 5
    },
    {
      "id": 45497,
      "title": "My single sub to get 70%",
      "comment_count": 18,
      "views": 0,
      "votes": 5
    },
    {
      "id": 45728,
      "title": "Did anyone try text detection/classification approaches?",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 45722,
      "title": "almost zero shakeup",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 45713,
      "title": "where is my result????",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 45715,
      "title": "Methods that should've worked (but didn't)",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44581,
      "title": "recipe for high accuracy model",
      "comment_count": 15,
      "views": 0,
      "votes": 26
    },
    {
      "id": 45391,
      "title": "'imagenet_utils' is not defined when reloading model to predict only",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 45208,
      "title": "Team Merger?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 44991,
      "title": "Let's team up",
      "comment_count": 5,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44789,
      "title": "Training with gradient accumulation doesn't work the same ",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 44366,
      "title": "Train/Val Split",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 44115,
      "title": "Google Cloud Storage Bucket",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 44303,
      "title": "Small batch size (8) and decreasing accuracy",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 43305,
      "title": "automated network design",
      "comment_count": 8,
      "views": 0,
      "votes": null
    },
    {
      "id": 44132,
      "title": "Has anyone trained hierarchy classification net in yolo9000 successfully?",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41652,
      "title": "Single model performance",
      "comment_count": 91,
      "views": 0,
      "votes": 43
    },
    {
      "id": 42069,
      "title": "repeating resnet101 LB=0.74 results",
      "comment_count": 22,
      "views": 0,
      "votes": 25
    },
    {
      "id": 43991,
      "title": "On updating regularizers of a pre-trained model",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43259,
      "title": "How do I access data remotely?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 43329,
      "title": "about this competition",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42755,
      "title": "Any loss for only top1 acc?",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 42067,
      "title": "Late starter",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 42951,
      "title": "test id's do not match",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42840,
      "title": "Has anyone tried to train a model from scratch?",
      "comment_count": 7,
      "views": 0,
      "votes": null
    },
    {
      "id": 41123,
      "title": "Classifying products with multiple images",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 42536,
      "title": "Unclean Data",
      "comment_count": 3,
      "views": 0,
      "votes": 4
    },
    {
      "id": 40412,
      "title": "Is there any free or less price, Platform-as-a-Service for training deep learning models in the cloud?",
      "comment_count": 4,
      "views": 0,
      "votes": 4
    },
    {
      "id": 42683,
      "title": "Prediction by a generator (Keras)",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 41632,
      "title": "any one trying multiple classifier approach?",
      "comment_count": 7,
      "views": 0,
      "votes": 2
    },
    {
      "id": 42589,
      "title": "1GB Training Data",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40320,
      "title": "How to feed data into net?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42283,
      "title": "parallel execution on spark cluster",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42261,
      "title": "what is the Convoluted predictions?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 39479,
      "title": "Can you we download train.bson by torrent? ",
      "comment_count": 10,
      "views": 0,
      "votes": 2
    },
    {
      "id": 42107,
      "title": "Trying to train inception_resnet_v2 on aws",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 41865,
      "title": "Does anyone use Chainer Deep Learning Framework?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41747,
      "title": "if this works ...\"Reinforcement Learning for Learning Rate Control\"",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 41650,
      "title": "Product ID and category_id are related.",
      "comment_count": 9,
      "views": 0,
      "votes": 9
    },
    {
      "id": 40715,
      "title": "Keras starter",
      "comment_count": 14,
      "views": 0,
      "votes": 18
    },
    {
      "id": 41506,
      "title": "Full functional deepsense.ai experiment example",
      "comment_count": 8,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41609,
      "title": "PyTorch MongoDB Dataloader ",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 41513,
      "title": "How to save images into my local disc or to save as output files in kernel?",
      "comment_count": 5,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41224,
      "title": "keras importance sampling",
      "comment_count": 3,
      "views": 0,
      "votes": 7
    },
    {
      "id": 41075,
      "title": "Preprocess input mode for pretrain Keras model",
      "comment_count": 5,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40503,
      "title": "Tensorflow Error: Cannot Fixed by Myself",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40922,
      "title": "Multi-gpu error:the accuracy of one gpu is much lower than two gpu",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 41030,
      "title": "Read bson file directly will be faster than read images?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 41374,
      "title": "Some help need to verify quality of my train.bson",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 39777,
      "title": "A few of the things I tried",
      "comment_count": 7,
      "views": 0,
      "votes": 18
    },
    {
      "id": 41027,
      "title": "Retraining a Network with Random Access?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41044,
      "title": "how to factorise trained convolution filters?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 41121,
      "title": "Using MXNet backend in Keras for model parallelism ",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 40934,
      "title": "Reusing model",
      "comment_count": 20,
      "views": 0,
      "votes": 10
    },
    {
      "id": 41176,
      "title": "Principal Component analysis useful?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40937,
      "title": "tricks",
      "comment_count": 4,
      "views": 0,
      "votes": 15
    },
    {
      "id": 40664,
      "title": "Single model, single crop Top LB score?",
      "comment_count": 35,
      "views": 0,
      "votes": 5
    },
    {
      "id": 40827,
      "title": "Caution for EXT4 file counts. And any advice?",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40886,
      "title": "Data augmentation methods in image classification",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 40033,
      "title": "What's your approach?",
      "comment_count": 17,
      "views": 0,
      "votes": 6
    },
    {
      "id": 40713,
      "title": "Pretrained keras models available for Kernels",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40693,
      "title": "single model prediction performance",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 39800,
      "title": "how long your model trained? ",
      "comment_count": 39,
      "views": 0,
      "votes": 1
    },
    {
      "id": 39925,
      "title": "Validation vs test scores?",
      "comment_count": 13,
      "views": 0,
      "votes": 8
    },
    {
      "id": 39776,
      "title": "auxillary loss for hierarchies + evaluation",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40178,
      "title": "Torrent Files",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40393,
      "title": "Can We use dataset for Academic research purpose ?",
      "comment_count": 1,
      "views": 0,
      "votes": 7
    },
    {
      "id": 40569,
      "title": "Dealing with JSON files instead of BSON",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40428,
      "title": "Create Train,Validate and Test Set",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40558,
      "title": "Use only bson package",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 39601,
      "title": "You nhave 72 hours to download",
      "comment_count": 12,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40265,
      "title": "Distribution of Test Dataset",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40083,
      "title": "Any tried to fit the training dataset??",
      "comment_count": 9,
      "views": 0,
      "votes": null
    },
    {
      "id": 40249,
      "title": "How to modify ResNet?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 39809,
      "title": "Does Kaggle have any plans to give free Google cloud GPU access like other competitions?",
      "comment_count": 9,
      "views": 0,
      "votes": 5
    },
    {
      "id": 39936,
      "title": "level3 category names are not unique",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 39728,
      "title": "Image Similarity Benchmark",
      "comment_count": 2,
      "views": 0,
      "votes": 9
    },
    {
      "id": 40006,
      "title": "InvalidBSON: objsize too large",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 39583,
      "title": "378 GB of images in training dataset",
      "comment_count": 20,
      "views": 0,
      "votes": 9
    },
    {
      "id": 40052,
      "title": "Could you please generate MD5 hashes for files ",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 39698,
      "title": "Question on Data Management",
      "comment_count": 13,
      "views": 0,
      "votes": 1
    },
    {
      "id": 39890,
      "title": "Anyone Interested in teaming up?",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 39700,
      "title": "Torrent or train split please.",
      "comment_count": 6,
      "views": 0,
      "votes": 4
    },
    {
      "id": 39466,
      "title": "any one willing to team up ",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 39859,
      "title": "Any success with pre-trained models?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 39780,
      "title": "Data augmentation",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 39509,
      "title": "so... how to put this data into memory?",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 39520,
      "title": "train.bson will download faster now",
      "comment_count": 1,
      "views": 0,
      "votes": 8
    },
    {
      "id": 39654,
      "title": "MD5 Please",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 39653,
      "title": "Data file Size",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 39472,
      "title": "Very Huge DataSets !! ",
      "comment_count": 8,
      "views": 0,
      "votes": 2
    },
    {
      "id": 39631,
      "title": "Inception v3",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 39489,
      "title": "How to wget the data  files ?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 39462,
      "title": "Example Kernel missing :(",
      "comment_count": 10,
      "views": 0,
      "votes": 2
    },
    {
      "id": 39490,
      "title": "Very slow downloading (train.bson)",
      "comment_count": 6,
      "views": 0,
      "votes": 4
    },
    {
      "id": 39492,
      "title": "How to download data directly in a cloud server (AWS/Google Cloud)",
      "comment_count": 3,
      "views": 0,
      "votes": 6
    },
    {
      "id": 39532,
      "title": "train_example.bson has only 82 products?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 39473,
      "title": "We have a fast paced ImageNet here ...",
      "comment_count": 1,
      "views": 0,
      "votes": null
    }
  ],
  "errors": []
}