{
  "total": 251,
  "topics": [
    {
      "id": 223195,
      "title": "Welcome to the Molecular Translation Competition!",
      "comment_count": 33,
      "views": 0,
      "votes": 60
    },
    {
      "id": 245391,
      "title": "Leaderboard Finalized - Congratulations to the Winners; Recap",
      "comment_count": 4,
      "views": 0,
      "votes": 13
    },
    {
      "id": 231318,
      "title": "Clarity about External Data use regarding PubChem, provided InChIs, etc.",
      "comment_count": 35,
      "views": 0,
      "votes": 33
    },
    {
      "id": 244266,
      "title": "18th place solution",
      "comment_count": 9,
      "views": 0,
      "votes": 19
    },
    {
      "id": 483801,
      "title": "Some work after winning 1st place in the competition",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 225856,
      "title": "Resnet + LSTM with attention starter",
      "comment_count": 49,
      "views": 0,
      "votes": 141
    },
    {
      "id": 243102,
      "title": "rdkit is more crucial",
      "comment_count": 35,
      "views": 0,
      "votes": 28
    },
    {
      "id": 223218,
      "title": "🎊 🎉 Research Papers in Deep Learning and Chemical Structures (Image Data)🌟 ✨",
      "comment_count": 31,
      "views": 0,
      "votes": 132
    },
    {
      "id": 231190,
      "title": "[completed] transformer starter kit ... in pytorch",
      "comment_count": 216,
      "views": 0,
      "votes": 179
    },
    {
      "id": 226245,
      "title": "Difficulties training a CNN",
      "comment_count": 11,
      "views": 0,
      "votes": 4
    },
    {
      "id": 223419,
      "title": "💥 💥   Image Captioning  - DL Research Papers + Code🔥🔥",
      "comment_count": 6,
      "views": 0,
      "votes": 46
    },
    {
      "id": 243114,
      "title": "On continuing research in this area",
      "comment_count": 5,
      "views": 0,
      "votes": 6
    },
    {
      "id": 243809,
      "title": "10th solution & thoughts - Part2. LB 0.61 [ Think out of the box version ]",
      "comment_count": 3,
      "views": 0,
      "votes": 40
    },
    {
      "id": 247472,
      "title": "Excited and curious to see 1st place solution ",
      "comment_count": 4,
      "views": 0,
      "votes": 5
    },
    {
      "id": 243932,
      "title": "3rd place solution: brief summary",
      "comment_count": 24,
      "views": 0,
      "votes": 79
    },
    {
      "id": 243779,
      "title": "7th Place Solution : My First Gold And I am in Tears",
      "comment_count": 45,
      "views": 0,
      "votes": 102
    },
    {
      "id": 243917,
      "title": "24th place (using molecular graph and atom coordinates)",
      "comment_count": 11,
      "views": 0,
      "votes": 35
    },
    {
      "id": 246322,
      "title": "About those A100s ...",
      "comment_count": 8,
      "views": 0,
      "votes": 21
    },
    {
      "id": 231065,
      "title": "Batched beam search implementation",
      "comment_count": 21,
      "views": 0,
      "votes": 76
    },
    {
      "id": 247661,
      "title": "27th place key highlights in 60 seconds",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 243824,
      "title": "12th place...",
      "comment_count": 27,
      "views": 0,
      "votes": 86
    },
    {
      "id": 243811,
      "title": "Congrats for the 2nd place, you're truly genius.",
      "comment_count": 39,
      "views": 0,
      "votes": 72
    },
    {
      "id": 244166,
      "title": "13th Place Team (a story from a Kaggle newbie's view)",
      "comment_count": 10,
      "views": 0,
      "votes": 30
    },
    {
      "id": 245704,
      "title": "6th Place Solution - General Overview",
      "comment_count": 0,
      "views": 0,
      "votes": 30
    },
    {
      "id": 244601,
      "title": "Ensemble directly on submissions",
      "comment_count": 2,
      "views": 0,
      "votes": 26
    },
    {
      "id": 243478,
      "title": "Attention is What You Get",
      "comment_count": 4,
      "views": 0,
      "votes": 5
    },
    {
      "id": 243845,
      "title": "47th Place Solution & Code",
      "comment_count": 8,
      "views": 0,
      "votes": 31
    },
    {
      "id": 243766,
      "title": "10th solution private 0.61 : TNT/VIT/CAIT ensemble + tricks",
      "comment_count": 14,
      "views": 0,
      "votes": 60
    },
    {
      "id": 244491,
      "title": "How to submit code and get bonus？",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 243943,
      "title": "5th Place Solution",
      "comment_count": 15,
      "views": 0,
      "votes": 53
    },
    {
      "id": 243787,
      "title": "4th Place Solution",
      "comment_count": 23,
      "views": 0,
      "votes": 59
    },
    {
      "id": 244078,
      "title": "8th Place Solution🔺",
      "comment_count": 15,
      "views": 0,
      "votes": 36
    },
    {
      "id": 243761,
      "title": "9th place: SWIN/VIT/CAIT + Heng decoder with code: tugstugi's part (updated)",
      "comment_count": 17,
      "views": 0,
      "votes": 45
    },
    {
      "id": 243819,
      "title": "about that share",
      "comment_count": 24,
      "views": 0,
      "votes": 22
    },
    {
      "id": 243868,
      "title": "70th place solution and code",
      "comment_count": 2,
      "views": 0,
      "votes": 21
    },
    {
      "id": 243820,
      "title": "50th Place Solution",
      "comment_count": 6,
      "views": 0,
      "votes": 19
    },
    {
      "id": 244031,
      "title": "9th Place Solution",
      "comment_count": 5,
      "views": 0,
      "votes": 36
    },
    {
      "id": 243792,
      "title": "Top Solutions and Approaches",
      "comment_count": 1,
      "views": 0,
      "votes": 16
    },
    {
      "id": 243951,
      "title": "15th adversarial validation etc",
      "comment_count": 2,
      "views": 0,
      "votes": 22
    },
    {
      "id": 243751,
      "title": "Just wanna say thanks to Cher Keng Heng",
      "comment_count": 3,
      "views": 0,
      "votes": 27
    },
    {
      "id": 244012,
      "title": "110th place effort",
      "comment_count": 2,
      "views": 0,
      "votes": 11
    },
    {
      "id": 243993,
      "title": "Sometimes a good harware can be a game changer",
      "comment_count": 3,
      "views": 0,
      "votes": 6
    },
    {
      "id": 243567,
      "title": "Final day be like...",
      "comment_count": 13,
      "views": 0,
      "votes": 21
    },
    {
      "id": 242592,
      "title": "Some insights and observations",
      "comment_count": 8,
      "views": 0,
      "votes": 19
    },
    {
      "id": 243679,
      "title": "Good Luck Everyone ....",
      "comment_count": 1,
      "views": 0,
      "votes": 11
    },
    {
      "id": 243683,
      "title": "shame on you",
      "comment_count": 3,
      "views": 0,
      "votes": 7
    },
    {
      "id": 243535,
      "title": "some good nights sleep  😄",
      "comment_count": 2,
      "views": 0,
      "votes": 10
    },
    {
      "id": 237284,
      "title": "Yet another paper that takes inspiration from Kaggle without giving credit",
      "comment_count": 14,
      "views": 0,
      "votes": 91
    },
    {
      "id": 243777,
      "title": "Special thanks ",
      "comment_count": 0,
      "views": 0,
      "votes": 9
    },
    {
      "id": 243789,
      "title": "Thank you to Kaggle team and the organisers!!",
      "comment_count": 0,
      "views": 0,
      "votes": 6
    },
    {
      "id": 243767,
      "title": "Interesting Experiments & Learnings",
      "comment_count": 0,
      "views": 0,
      "votes": 6
    },
    {
      "id": 243782,
      "title": "why I can't see my team on the leaderboard?",
      "comment_count": 12,
      "views": 0,
      "votes": 3
    },
    {
      "id": 242403,
      "title": "One Day Deadline Extension",
      "comment_count": 21,
      "views": 0,
      "votes": 8
    },
    {
      "id": 242794,
      "title": "Deleted user in medal zone",
      "comment_count": 13,
      "views": 0,
      "votes": 2
    },
    {
      "id": 241716,
      "title": "Transformer Decoder + EfficientNetV2 Encoder – Full Pipeline – [3.71 CV]",
      "comment_count": 15,
      "views": 0,
      "votes": 28
    },
    {
      "id": 243472,
      "title": "A day to go?",
      "comment_count": 5,
      "views": 0,
      "votes": 3
    },
    {
      "id": 242582,
      "title": "Tokenizer face palm",
      "comment_count": 17,
      "views": 0,
      "votes": 12
    },
    {
      "id": 243131,
      "title": "a vision transformer basline",
      "comment_count": 6,
      "views": 0,
      "votes": 3
    },
    {
      "id": 242082,
      "title": "Ensembling Techniques",
      "comment_count": 14,
      "views": 0,
      "votes": 6
    },
    {
      "id": 242778,
      "title": "PseudoLabelling",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 243284,
      "title": "TPU socket closed in the middle of committed notebook running.",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 242179,
      "title": "The Could not find evaluation metric \"\" error",
      "comment_count": 8,
      "views": 0,
      "votes": 8
    },
    {
      "id": 242200,
      "title": "Evaluation Exception: Missing submission schema mapping",
      "comment_count": 8,
      "views": 0,
      "votes": 7
    },
    {
      "id": 241325,
      "title": "Multitask learning based on various components of InChI",
      "comment_count": 8,
      "views": 0,
      "votes": null
    },
    {
      "id": 239236,
      "title": "Pretrained Weights and code for Efficientnet V2 Models",
      "comment_count": 27,
      "views": 0,
      "votes": 10
    },
    {
      "id": 242246,
      "title": "TTA: Test time augmentations for better score.",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 240829,
      "title": "GPU Crisis",
      "comment_count": 13,
      "views": 0,
      "votes": 22
    },
    {
      "id": 241155,
      "title": "Has ensembling helped any one ?",
      "comment_count": 2,
      "views": 0,
      "votes": 5
    },
    {
      "id": 240624,
      "title": "Let's Start the Memes thread :)",
      "comment_count": 6,
      "views": 0,
      "votes": 16
    },
    {
      "id": 239350,
      "title": "Four Grandmasters on Top Team is Anti-Competitive",
      "comment_count": 54,
      "views": 0,
      "votes": 12
    },
    {
      "id": 239595,
      "title": "EfficientNetV2 Encoder Implemented [4.97 LB] – Training+Inference In Under 3 Hours 🔥🔥🔥",
      "comment_count": 11,
      "views": 0,
      "votes": 46
    },
    {
      "id": 242354,
      "title": "Molecular Translation using Visual attention",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 241557,
      "title": "Need some guide in building a simple transformer Model baseline",
      "comment_count": 2,
      "views": 0,
      "votes": 5
    },
    {
      "id": 238546,
      "title": "FP16, transformer training, NaN",
      "comment_count": 15,
      "views": 0,
      "votes": 11
    },
    {
      "id": 239750,
      "title": "Colab version TPU efficientNet v2 pipeline available",
      "comment_count": 10,
      "views": 0,
      "votes": 7
    },
    {
      "id": 240282,
      "title": "Found A Bug in YNakama's Tokenizer",
      "comment_count": 8,
      "views": 0,
      "votes": 32
    },
    {
      "id": 241059,
      "title": "Rotation Detector Model Architecture",
      "comment_count": 0,
      "views": 0,
      "votes": 6
    },
    {
      "id": 240862,
      "title": "My Poor Orientation Model",
      "comment_count": 0,
      "views": 0,
      "votes": 9
    },
    {
      "id": 241277,
      "title": "Optimal Image Size",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 241602,
      "title": "Is it required to name the csv file \"submission.csv\" ?",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 241101,
      "title": "Any faster way to move files from google collab environment to kaggle dataset? ",
      "comment_count": 2,
      "views": 0,
      "votes": 2
    },
    {
      "id": 241399,
      "title": "Would Kaggle change to save partial result instead of cancelling all after timeout?",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 236125,
      "title": "Any idea about ensemble?  ",
      "comment_count": 29,
      "views": 0,
      "votes": 17
    },
    {
      "id": 239558,
      "title": "CV LB Gap?",
      "comment_count": 16,
      "views": 0,
      "votes": 6
    },
    {
      "id": 241380,
      "title": "Looking for a team ",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 240586,
      "title": "Tuning on Longer Lengths",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 240029,
      "title": "Batched Beam Search in TensorFlow?",
      "comment_count": 4,
      "views": 0,
      "votes": 4
    },
    {
      "id": 240474,
      "title": "Wow! Gold < 1.0",
      "comment_count": 2,
      "views": 0,
      "votes": 4
    },
    {
      "id": 240316,
      "title": "LAMB - \"Large Batch Optimization for Deep Learning: Training BERT in 76 minutes\"",
      "comment_count": 1,
      "views": 0,
      "votes": 6
    },
    {
      "id": 239906,
      "title": "How can beam search be used on the basis of fairseq?",
      "comment_count": 2,
      "views": 0,
      "votes": 6
    },
    {
      "id": 240088,
      "title": "Uncommon topic: Best single model (for rotations)",
      "comment_count": 3,
      "views": 0,
      "votes": 5
    },
    {
      "id": 240179,
      "title": "Anyone using Dacon.ai top solutions.",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 239691,
      "title": "CV-LB error",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 238551,
      "title": "Basic PP that maybe can help",
      "comment_count": 5,
      "views": 0,
      "votes": 12
    },
    {
      "id": 238989,
      "title": "what is the difference between Kaggle TPU VM and Colab TPU VM",
      "comment_count": 4,
      "views": 0,
      "votes": 5
    },
    {
      "id": 238494,
      "title": "Symmetry is now reached!",
      "comment_count": 3,
      "views": 0,
      "votes": 29
    },
    {
      "id": 239673,
      "title": "ChemBERTa ",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 239590,
      "title": "Transformer Embedding Dim",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 238475,
      "title": "Why is LB consistently higher than CV?",
      "comment_count": 5,
      "views": 0,
      "votes": 2
    },
    {
      "id": 235509,
      "title": "Understanding Vocabulary Embedding Dimension",
      "comment_count": 6,
      "views": 0,
      "votes": 6
    },
    {
      "id": 238568,
      "title": "deepmind paper suggest that transformer can learn to sketch molecular svg image?",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 237645,
      "title": "Friendly Reminder Showing Additional EfficientNET Keras/TF Resources",
      "comment_count": 0,
      "views": 0,
      "votes": 8
    },
    {
      "id": 234099,
      "title": "What Aspect Ratio Are You Using As Model Input And Why?",
      "comment_count": 34,
      "views": 0,
      "votes": 14
    },
    {
      "id": 233991,
      "title": "Had a pretty hard time downloading the data ",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 238197,
      "title": "I can not start my kaggle notebook for one day.",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 235971,
      "title": "we found bumblebee and some transformers ...",
      "comment_count": 14,
      "views": 0,
      "votes": 36
    },
    {
      "id": 234718,
      "title": "how to train using test images : self-supervised tricks that every kaggler should know",
      "comment_count": 5,
      "views": 0,
      "votes": 35
    },
    {
      "id": 229978,
      "title": "Competition Unfairness (fixed)",
      "comment_count": 39,
      "views": 0,
      "votes": 94
    },
    {
      "id": 223174,
      "title": "New to Kaggle or Machine Learning? Come Say Hi!",
      "comment_count": 18,
      "views": 0,
      "votes": 20
    },
    {
      "id": 229258,
      "title": "Rdkit crashes when translate an InChI string I generated to molecule ",
      "comment_count": 4,
      "views": 0,
      "votes": 5
    },
    {
      "id": 236494,
      "title": "np.save() HDD space full!",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 236298,
      "title": "broken pixel chemical structural diagram",
      "comment_count": 3,
      "views": 0,
      "votes": 4
    },
    {
      "id": 236166,
      "title": "Problem with predicted InChI's /c layer text strings",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 230887,
      "title": "we are watching a race of gpu power",
      "comment_count": 35,
      "views": 0,
      "votes": 46
    },
    {
      "id": 236017,
      "title": "better encoder?",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 234680,
      "title": "Sharing some results",
      "comment_count": 8,
      "views": 0,
      "votes": 43
    },
    {
      "id": 229720,
      "title": "some model suggestions",
      "comment_count": 42,
      "views": 0,
      "votes": 43
    },
    {
      "id": 228975,
      "title": "how to improve inference speed??",
      "comment_count": 6,
      "views": 0,
      "votes": 4
    },
    {
      "id": 234553,
      "title": "4h training on TPU for LB 5.3: good or inefficient? (LTSM+ATT)",
      "comment_count": 9,
      "views": 0,
      "votes": 7
    },
    {
      "id": 233389,
      "title": "Distributed Training and Inference on TPU",
      "comment_count": 9,
      "views": 0,
      "votes": 11
    },
    {
      "id": 232685,
      "title": "Improve score of already trained models!",
      "comment_count": 20,
      "views": 0,
      "votes": 31
    },
    {
      "id": 235354,
      "title": "sequential decoder evaluation takes ages ",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 230376,
      "title": "How to write efficient PyTorch code on TPU 🔥🔥",
      "comment_count": 9,
      "views": 0,
      "votes": 16
    },
    {
      "id": 234857,
      "title": "How is the labelling order determined for InChI?",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 224257,
      "title": "Best Single Model",
      "comment_count": 235,
      "views": 0,
      "votes": 108
    },
    {
      "id": 233672,
      "title": "Compliments To \"Eight V100s are Working Hard\" Getting Under 1.00!!",
      "comment_count": 20,
      "views": 0,
      "votes": 11
    },
    {
      "id": 234461,
      "title": "Strategy / Hybrid method  using logical process and deep learning / Object detection + CNN + rdkit",
      "comment_count": 2,
      "views": 0,
      "votes": 4
    },
    {
      "id": 234729,
      "title": "TPU training with PyTorch/Lightning",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 233927,
      "title": "Information encoded in the image margins",
      "comment_count": 9,
      "views": 0,
      "votes": 21
    },
    {
      "id": 234462,
      "title": "How to train model",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 228610,
      "title": "Are you using all the training data at this time?",
      "comment_count": 12,
      "views": 0,
      "votes": 28
    },
    {
      "id": 227513,
      "title": "A note on LR schedulers",
      "comment_count": 7,
      "views": 0,
      "votes": 17
    },
    {
      "id": 231800,
      "title": "How everyone is training such a big dataset within the kaggle timelimit?",
      "comment_count": 9,
      "views": 0,
      "votes": 6
    },
    {
      "id": 233387,
      "title": "there is something fishy about image size",
      "comment_count": 25,
      "views": 0,
      "votes": 18
    },
    {
      "id": 232380,
      "title": "if i want to build a inchi-to-image generator, what paper can i follows?",
      "comment_count": 8,
      "views": 0,
      "votes": 3
    },
    {
      "id": 233834,
      "title": "Width/height ratios of enc input res around 2 most 'efficient'",
      "comment_count": 0,
      "views": 0,
      "votes": 12
    },
    {
      "id": 232333,
      "title": "EfficientNet V2",
      "comment_count": 3,
      "views": 0,
      "votes": 11
    },
    {
      "id": 233561,
      "title": "Trouble training with pytorch",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 223175,
      "title": "Looking for a Team Megathread",
      "comment_count": 61,
      "views": 0,
      "votes": 13
    },
    {
      "id": 231950,
      "title": "EfficientNet + LSTM - LB: 8.89 (with credit to Y. Nakama, duh ;-)",
      "comment_count": 5,
      "views": 0,
      "votes": 28
    },
    {
      "id": 232025,
      "title": "Sharing Mean and Standard deviation of images",
      "comment_count": 8,
      "views": 0,
      "votes": 6
    },
    {
      "id": 231400,
      "title": "Making inferencing faster",
      "comment_count": 17,
      "views": 0,
      "votes": 26
    },
    {
      "id": 232804,
      "title": "How To Calculate Metrics Accurately?",
      "comment_count": 27,
      "views": 0,
      "votes": 5
    },
    {
      "id": 231961,
      "title": "Pytorch + Numpy bug that you should be aware. ",
      "comment_count": 7,
      "views": 0,
      "votes": 26
    },
    {
      "id": 229640,
      "title": "non autoregressive decoder?",
      "comment_count": 13,
      "views": 0,
      "votes": 9
    },
    {
      "id": 232619,
      "title": "How to unzip data efficiently ?",
      "comment_count": 7,
      "views": 0,
      "votes": 6
    },
    {
      "id": 232878,
      "title": "Private Test Set",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 232486,
      "title": "How InChi really works here",
      "comment_count": 2,
      "views": 0,
      "votes": 4
    },
    {
      "id": 232354,
      "title": "LG Smiles solutions",
      "comment_count": 0,
      "views": 0,
      "votes": 11
    },
    {
      "id": 232413,
      "title": "Dynamic Minibatch Trimming to speed up training and inference.",
      "comment_count": 0,
      "views": 0,
      "votes": 8
    },
    {
      "id": 229984,
      "title": "Dacon Top Solution Outline",
      "comment_count": 23,
      "views": 0,
      "votes": 43
    },
    {
      "id": 231172,
      "title": "Effect of image sizes",
      "comment_count": 33,
      "views": 0,
      "votes": 41
    },
    {
      "id": 232027,
      "title": "Dataset - 404 Error",
      "comment_count": 5,
      "views": 0,
      "votes": 8
    },
    {
      "id": 232051,
      "title": "sota: from vgg to resnet to efficient-net, then back to resnet, back to vgg again",
      "comment_count": 2,
      "views": 0,
      "votes": 15
    },
    {
      "id": 232401,
      "title": "Possible Way To Speed Up Training - Idea ",
      "comment_count": 0,
      "views": 0,
      "votes": 5
    },
    {
      "id": 231739,
      "title": "Data Download Returning 404",
      "comment_count": 8,
      "views": 0,
      "votes": 7
    },
    {
      "id": 231470,
      "title": "Download Link Broken? 404 on Download All",
      "comment_count": 9,
      "views": 0,
      "votes": 11
    },
    {
      "id": 232087,
      "title": "extra_approved_InChIs",
      "comment_count": 3,
      "views": 0,
      "votes": 4
    },
    {
      "id": 223423,
      "title": "Here's why ihelon's image_id_2_path function is genius (and might save you 20 min)",
      "comment_count": 5,
      "views": 0,
      "votes": 60
    },
    {
      "id": 232158,
      "title": "Has someone meet Exposure bias problem?",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 232538,
      "title": "The case of a misshared top3 solutions",
      "comment_count": 1,
      "views": 0,
      "votes": -10
    },
    {
      "id": 231871,
      "title": "Resized Images",
      "comment_count": 0,
      "views": 0,
      "votes": 7
    },
    {
      "id": 230095,
      "title": "Computing Levenshtein Distance Easily",
      "comment_count": 2,
      "views": 0,
      "votes": 7
    },
    {
      "id": 230498,
      "title": "Tips for Transformer inference",
      "comment_count": 9,
      "views": 0,
      "votes": 59
    },
    {
      "id": 231035,
      "title": "384x384 - Raw TFRecords [+ Easily Create More]",
      "comment_count": 3,
      "views": 0,
      "votes": 10
    },
    {
      "id": 231330,
      "title": "Confusion about masked attention in the MultiHeadAttention (pytorch).",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 229279,
      "title": "Question about copyright issues",
      "comment_count": 46,
      "views": 0,
      "votes": 26
    },
    {
      "id": 230528,
      "title": "Public/Private split",
      "comment_count": 7,
      "views": 0,
      "votes": 6
    },
    {
      "id": 231500,
      "title": "Is it possible to train with Pytorch Xla (TPU)?  ",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 230745,
      "title": "What approch should take if I want to use small version of Dataset for Training?",
      "comment_count": 6,
      "views": 0,
      "votes": 5
    },
    {
      "id": 230473,
      "title": "Stereochemical layer confusion (and clarification)",
      "comment_count": 4,
      "views": 0,
      "votes": 8
    },
    {
      "id": 223223,
      "title": "💥💥Github Resources + Pre-Trained Models for Deep Learning + Chemical Structures 🔥🔥",
      "comment_count": 7,
      "views": 0,
      "votes": 59
    },
    {
      "id": 230185,
      "title": "Generated InCHI does not need to be valid",
      "comment_count": 28,
      "views": 0,
      "votes": 6
    },
    {
      "id": 228220,
      "title": "My model predictions are better than training labels",
      "comment_count": 19,
      "views": 0,
      "votes": 44
    },
    {
      "id": 229243,
      "title": "How are atoms numerated? And can transformers learn to numerate them?",
      "comment_count": 9,
      "views": 0,
      "votes": 7
    },
    {
      "id": 230416,
      "title": "Reproducibility of top submissions",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 224104,
      "title": "Canonicalization Quirks in Train Data",
      "comment_count": 3,
      "views": 0,
      "votes": 14
    },
    {
      "id": 229578,
      "title": "Normalize your predictions",
      "comment_count": 6,
      "views": 0,
      "votes": 21
    },
    {
      "id": 230170,
      "title": "transformer(decoder)+resnet baseline 55+",
      "comment_count": 0,
      "views": 0,
      "votes": 6
    },
    {
      "id": 229557,
      "title": "MADGRAD: A New Optimizer to Try",
      "comment_count": 1,
      "views": 0,
      "votes": 15
    },
    {
      "id": 230162,
      "title": "could Kaggle provide mini dataset for code debugging and big dataset for competition?",
      "comment_count": 2,
      "views": 0,
      "votes": 2
    },
    {
      "id": 229816,
      "title": "Img2Mol - Accurate SMILES Recognition from Molecular Graphical Depictions",
      "comment_count": 1,
      "views": 0,
      "votes": 7
    },
    {
      "id": 230134,
      "title": "OOM error when using torch_xla",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 229992,
      "title": "Help With TF2 Image Captioning Pipeline",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 230145,
      "title": "Confused about preprocess part 😞 ! ",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 229874,
      "title": "Dataset Storage when Using PyTorch on TPU",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 228139,
      "title": "Chemical Formula Prediction as Multi-Output Regression",
      "comment_count": 8,
      "views": 0,
      "votes": 22
    },
    {
      "id": 229459,
      "title": "TPU Socket closed ERROR",
      "comment_count": 3,
      "views": 0,
      "votes": 6
    },
    {
      "id": 229168,
      "title": "Stereochemical Layer (/t) Prediction (93% Accuracy)",
      "comment_count": 2,
      "views": 0,
      "votes": 13
    },
    {
      "id": 229515,
      "title": "Test Images manipulation?",
      "comment_count": 9,
      "views": 0,
      "votes": 2
    },
    {
      "id": 225876,
      "title": "PubChem molecule structure generator (InChI to PNG)",
      "comment_count": 4,
      "views": 0,
      "votes": 12
    },
    {
      "id": 227864,
      "title": "How much does it take you to build the first working pipeline?",
      "comment_count": 14,
      "views": 0,
      "votes": 11
    },
    {
      "id": 223466,
      "title": "Brief summary of \"How to calculate Levenshtein distance\"🔰",
      "comment_count": 12,
      "views": 0,
      "votes": 36
    },
    {
      "id": 226216,
      "title": "Use of pytesseract (OCR) to get letters ?",
      "comment_count": 4,
      "views": 0,
      "votes": 3
    },
    {
      "id": 228277,
      "title": "images to smile and smiles to InchI",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 223381,
      "title": "Winning solutions from similar competition: Molecular Translation into SMILES",
      "comment_count": 29,
      "views": 0,
      "votes": 100
    },
    {
      "id": 228511,
      "title": "How to use multi-gpus properly?",
      "comment_count": 4,
      "views": 0,
      "votes": 6
    },
    {
      "id": 224375,
      "title": "Understanding InChI (sublayer \"/c\")",
      "comment_count": 8,
      "views": 0,
      "votes": 50
    },
    {
      "id": 227458,
      "title": "Why TPU allotation process crashes...?",
      "comment_count": 2,
      "views": 0,
      "votes": 7
    },
    {
      "id": 223212,
      "title": "End-to-end transformers approach with ViT and vanilla transformer decoder (idea)",
      "comment_count": 18,
      "views": 0,
      "votes": 24
    },
    {
      "id": 227753,
      "title": "Using ResNet for encoding and Transformer for decoding",
      "comment_count": 6,
      "views": 0,
      "votes": 10
    },
    {
      "id": 223220,
      "title": "i think synthetic data generation + transformer will rule this competition",
      "comment_count": 4,
      "views": 0,
      "votes": 31
    },
    {
      "id": 228130,
      "title": "Ph.D opportunities in ML and Chemistry",
      "comment_count": 1,
      "views": 0,
      "votes": 5
    },
    {
      "id": 226024,
      "title": "The Plagiarism Begin",
      "comment_count": 21,
      "views": 0,
      "votes": 26
    },
    {
      "id": 225590,
      "title": "Concerns about the BMS dataset",
      "comment_count": 14,
      "views": 0,
      "votes": 42
    },
    {
      "id": 228269,
      "title": "download datasets",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 227244,
      "title": "Suitable (pretrained) CNNs for image captioning of molecules?",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 226292,
      "title": "how to simplify the problem",
      "comment_count": 6,
      "views": 0,
      "votes": 17
    },
    {
      "id": 224394,
      "title": "Problem with the evaluation metric",
      "comment_count": 14,
      "views": 0,
      "votes": 35
    },
    {
      "id": 226490,
      "title": "A request to show more decimal places on the leaderboard",
      "comment_count": 1,
      "views": 0,
      "votes": 9
    },
    {
      "id": 223239,
      "title": "How to solve this problem",
      "comment_count": 17,
      "views": 0,
      "votes": 47
    },
    {
      "id": 224294,
      "title": "Is manual annotation of training set allowed?",
      "comment_count": 8,
      "views": 0,
      "votes": 4
    },
    {
      "id": 226843,
      "title": "Error in using TPU",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 225943,
      "title": "Precise definition for InChI",
      "comment_count": 2,
      "views": 0,
      "votes": 6
    },
    {
      "id": 223233,
      "title": "Papers on optical chemical structure recognition",
      "comment_count": 2,
      "views": 0,
      "votes": 22
    },
    {
      "id": 223706,
      "title": "InChI to SMILES? A good Idea?",
      "comment_count": 10,
      "views": 0,
      "votes": 14
    },
    {
      "id": 225790,
      "title": "Formulas validation",
      "comment_count": 2,
      "views": 0,
      "votes": 7
    },
    {
      "id": 225519,
      "title": "What's Up With This Image ?? [Master Thread]",
      "comment_count": 6,
      "views": 0,
      "votes": 7
    },
    {
      "id": 223612,
      "title": "Extracting data - taking forever",
      "comment_count": 16,
      "views": 0,
      "votes": 6
    },
    {
      "id": 225861,
      "title": "How to generate additional data",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 224176,
      "title": "💥💥Transformers for Vision - Curated Resource Collection 🔥🔥",
      "comment_count": 2,
      "views": 0,
      "votes": 24
    },
    {
      "id": 223194,
      "title": "DECIMER: towards deep learning for chemical image recognition - ¿A good start?",
      "comment_count": 8,
      "views": 0,
      "votes": 26
    },
    {
      "id": 223457,
      "title": "[Paper]: About InChI chemical structure",
      "comment_count": 2,
      "views": 0,
      "votes": 9
    },
    {
      "id": 223503,
      "title": "Target InChI Notation Explained With Examples , And some pointers about Problem !",
      "comment_count": 7,
      "views": 0,
      "votes": 32
    },
    {
      "id": 225130,
      "title": "Understanding each layer of InChI, the numbering order of atoms",
      "comment_count": 0,
      "views": 0,
      "votes": 7
    },
    {
      "id": 225358,
      "title": "DECIMER from Images to SMILES",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 224516,
      "title": "Test is slightly different",
      "comment_count": 6,
      "views": 0,
      "votes": 9
    },
    {
      "id": 223471,
      "title": "My idea to address this problem",
      "comment_count": 6,
      "views": 0,
      "votes": 16
    },
    {
      "id": 224275,
      "title": "Attention: Errors in the dataset.",
      "comment_count": 11,
      "views": 0,
      "votes": 8
    },
    {
      "id": 223605,
      "title": "All you wanted to know about Molecular Translation and you were too afraid to ask",
      "comment_count": 2,
      "views": 0,
      "votes": 29
    },
    {
      "id": 223305,
      "title": "What on Earth is InChI? [4-minute video]",
      "comment_count": 10,
      "views": 0,
      "votes": 15
    },
    {
      "id": 223293,
      "title": "Encoder - Decoder for Translation",
      "comment_count": 2,
      "views": 0,
      "votes": 27
    },
    {
      "id": 224420,
      "title": "Do image IDs have any significance?",
      "comment_count": 3,
      "views": 0,
      "votes": 5
    },
    {
      "id": 223438,
      "title": "How to augment noise in images?",
      "comment_count": 4,
      "views": 0,
      "votes": 7
    },
    {
      "id": 224232,
      "title": "Optimizing Levenshtein distance doubt",
      "comment_count": 3,
      "views": 0,
      "votes": 6
    },
    {
      "id": 223779,
      "title": "Constant solutions",
      "comment_count": 10,
      "views": 0,
      "votes": 10
    },
    {
      "id": 224504,
      "title": "Publication on Image2Smiles",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 224443,
      "title": "Detailed explanation of InChI",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 223230,
      "title": "Low quality of images",
      "comment_count": 12,
      "views": 0,
      "votes": 9
    },
    {
      "id": 224350,
      "title": "Preprocessing and padding",
      "comment_count": 11,
      "views": 0,
      "votes": 4
    },
    {
      "id": 223477,
      "title": "Resized and reshaped datasets - 128/256px, JPG/PNG, padded/original ratio",
      "comment_count": 6,
      "views": 0,
      "votes": 29
    },
    {
      "id": 223213,
      "title": "To the Host: Why InChl?",
      "comment_count": 9,
      "views": 0,
      "votes": 6
    },
    {
      "id": 224100,
      "title": "Impact of Synthetic Data",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 223269,
      "title": "Using SiameseNet is OK?",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 223464,
      "title": "find teams(找隊伍)",
      "comment_count": 2,
      "views": 0,
      "votes": 2
    },
    {
      "id": 223391,
      "title": "Resources for understanding Levenshtein distance ",
      "comment_count": 2,
      "views": 0,
      "votes": 8
    },
    {
      "id": 223184,
      "title": "InChI and Images on Wikidata ",
      "comment_count": 0,
      "views": 0,
      "votes": 22
    },
    {
      "id": 223400,
      "title": "[paper] Molecular Structure Extraction From DocumentsUsing Deep Learning",
      "comment_count": 1,
      "views": 0,
      "votes": 13
    },
    {
      "id": 223480,
      "title": "Commas in InChI column",
      "comment_count": 1,
      "views": 0,
      "votes": 5
    },
    {
      "id": 223378,
      "title": "Any interesting libraries that might be used?",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 223459,
      "title": "Question about license requirement",
      "comment_count": 1,
      "views": 0,
      "votes": 4
    }
  ],
  "errors": []
}