Per the Competition Rules, freely and publicly available external data is permitted in this competition, but must be posted to this forum thread no later than the Entry Deadline (one week before competition close).
\n\nOnce someone posts an external dataset to this thread, you do not need to re-post it if you are using the same one.
", "messages": [ { "id": "660248", "postDate": "10/28/2019 23:39:44", "content": "Per the Competition Rules, freely and publicly available external data is permitted in this competition, but must be posted to this forum thread no later than the Entry Deadline (one week before competition close).
\n\nOnce someone posts an external dataset to this thread, you do not need to re-post it if you are using the same one.
", "rawMarkdown": "Per the [Competition Rules](https://www.kaggle.com/c/tensorflow2-question-answering/rules), freely and publicly available external data is permitted in this competition, but must be posted to this forum thread no later than the Entry Deadline (one week before competition close).\n\nOnce someone posts an external dataset to this thread, you do not need to re-post it if you are using the same one.", "votes": null }, { "id": "664080", "postDate": "11/03/2019 05:55:57", "content": "Do we need to use tensorflow 2.0 or we are free to use any other framework?\nAlso how we can upload the trained model?
", "rawMarkdown": "Do we need to use tensorflow 2.0 or we are free to use any other framework?\nAlso how we can upload the trained model?", "votes": null }, { "id": "665122", "postDate": "11/04/2019 16:46:58", "content": "You are free to use any framework you choose. There are, however, a special set of prizes specific to use of TF2.0, which you can only be eligible for if you meet those requirements.
", "rawMarkdown": "You are free to use any framework you choose. There are, however, a [special set of prizes specific to use of TF2.0](https://www.kaggle.com/c/tensorflow2-question-answering/overview/prizes), which you can only be eligible for if you meet those requirements.", "votes": null }, { "id": "675311", "postDate": "11/17/2019 22:19:32", "content": "Do you consider pretrained models to be external data?
", "rawMarkdown": "Do you consider pretrained models to be external data?", "votes": null }, { "id": "681202", "postDate": "11/25/2019 20:32:02", "content": "Hello,\nI decided to enter this competition to freshen up my Python and ML skills. I thought live streaming my programming/modelling sessions would provide additional motivation to follow through to that goal. So as per public code sharing rules, I have to share the live streaming session link:
\n\n\n\nThis will be a semi-regular stream, but I plan on streaming the entire effort.
\n\nDisclaimer: due to the emphemeral nature of live video streaming, the videos may or may not disappear in the future. If you wish to save the broadcast, I suggest to use software made for that purpose.
", "rawMarkdown": "Hello,\nI decided to enter this competition to freshen up my Python and ML skills. I thought live streaming my programming/modelling sessions would provide additional motivation to follow through to that goal. So as per public code sharing rules, I have to share the live streaming session link:\n\nhttps://www.twitch.tv/cythie\n\nThis will be a semi-regular stream, but I plan on streaming the entire effort.\n\nDisclaimer: due to the emphemeral nature of live video streaming, the videos may or may not disappear in the future. If you wish to save the broadcast, I suggest to use software made for that purpose.", "votes": null }, { "id": "685504", "postDate": "12/01/2019 19:54:17", "content": "What's the \"No custom packages enabled in kernels\" means ? Can I use \"sys.path.append(xxx)\" ?
", "rawMarkdown": "What's the \"No custom packages enabled in kernels\" means ? Can I use \"sys.path.append(xxx)\" ?", "votes": null }, { "id": "696143", "postDate": "12/16/2019 07:13:20", "content": "one question: is the 3h limit of GPU counted only for the 88% private data or for the 100% whole data (public + private test data) ?
", "rawMarkdown": "one question: is the 3h limit of GPU counted only for the 88% private data or for the 100% whole data (public + private test data) ?", "votes": null }, { "id": "703214", "postDate": "12/25/2019 19:52:23", "content": "https://github.com/huggingface/transformers
", "rawMarkdown": "https://github.com/huggingface/transformers", "votes": null }, { "id": "706470", "postDate": "12/30/2019 12:43:05", "content": "All the models in https://github.com/huggingface/transformers
", "rawMarkdown": "All the models in https://github.com/huggingface/transformers", "votes": null }, { "id": "709925", "postDate": "01/04/2020 04:50:21", "content": "https://github.com/kamalkraj/ALBERT-TF2.0
", "rawMarkdown": "https://github.com/kamalkraj/ALBERT-TF2.0", "votes": null }, { "id": "711498", "postDate": "01/06/2020 07:31:38", "content": "https://nlp.cs.washington.edu/triviaqa/
", "rawMarkdown": "https://nlp.cs.washington.edu/triviaqa/", "votes": null }, { "id": "714594", "postDate": "01/09/2020 15:25:51", "content": "SQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers
", "rawMarkdown": "SQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers", "votes": null }, { "id": "715985", "postDate": "01/11/2020 05:08:30", "content": "Used or might be used:\nSQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers\nhttps://github.com/bojone/bert4keras\nhttps://github.com/CyberZHG/keras-bert\nhttps://ai.google.com/research/NaturalQuestions\nhttps://github.com/google-research/language\nand other datasets announced by other competitors of this match.
", "rawMarkdown": "Used or might be used:\nSQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers\nhttps://github.com/bojone/bert4keras\nhttps://github.com/CyberZHG/keras-bert\nhttps://ai.google.com/research/NaturalQuestions\nhttps://github.com/google-research/language\nand other datasets announced by other competitors of this match.", "votes": null }, { "id": "716015", "postDate": "01/11/2020 06:33:18", "content": "https://github.com/ThilinaRajapakse/simpletransformers
", "rawMarkdown": "https://github.com/ThilinaRajapakse/simpletransformers", "votes": null }, { "id": "717902", "postDate": "01/13/2020 18:57:10", "content": "https://www.kaggle.com/prokaj/bert-joint-baseline
\n\nhttps://www.kaggle.com/mmmarchetti/bert-joint
\n\ncopy of some python files from pip install sentencepiece bert-for-tf2
https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/1
\n\nmaybe I will try something from:
\n\n\n\n\n\nhttps://github.com/google-research/bert
\n\nhttps://github.com/google-research/ALBERT
\n\nhttps://github.com/nshepperd/gpt-2
", "rawMarkdown": "https://www.kaggle.com/prokaj/bert-joint-baseline\n\nhttps://www.kaggle.com/mmmarchetti/bert-joint\n\ncopy of some python files from ```pip install sentencepiece bert-for-tf2```\n\nhttps://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/1\n\nmaybe I will try something from:\n\nhttps://tfhub.dev/s?q=bert\n\nhttps://tfhub.dev/s?q=albert\n\nhttps://github.com/google-research/bert\n\nhttps://github.com/google-research/ALBERT\n\nhttps://github.com/nshepperd/gpt-2", "votes": null }, { "id": "719024", "postDate": "01/15/2020 03:22:43", "content": "I may or may not use these models / external datasets ヽ( ˊᵕˋ )ノ\nhttps://github.com/pytorch/fairseq/tree/master/examples/roberta\nhttps://github.com/pytorch/fairseq/tree/master/examples/bart\nhttps://github.com/zihangdai/xlnet\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttps://github.com/PaddlePaddle/ERNIE\nhttps://github.com/NervanaSystems/nlp-architect\nhttps://github.com/huawei-noah/Pretrained-Language-Model/tree/master/TinyBERT\nhttps://github.com/cooelf/SemBERT\nhttps://github.com/mandarjoshi90/coref\nhttps://github.com/facebookresearch/SpanBERT\nhttps://github.com/alexa/wqa_tanda\nhttps://github.com/mrqa/MRQA-Shared-Task-2019
", "rawMarkdown": "I may or may not use these models / external datasets ヽ( ˊᵕˋ )ノ\nhttps://github.com/pytorch/fairseq/tree/master/examples/roberta\nhttps://github.com/pytorch/fairseq/tree/master/examples/bart\nhttps://github.com/zihangdai/xlnet\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttps://github.com/PaddlePaddle/ERNIE\nhttps://github.com/NervanaSystems/nlp-architect\nhttps://github.com/huawei-noah/Pretrained-Language-Model/tree/master/TinyBERT\nhttps://github.com/cooelf/SemBERT\nhttps://github.com/mandarjoshi90/coref\nhttps://github.com/facebookresearch/SpanBERT\nhttps://github.com/alexa/wqa_tanda\nhttps://github.com/mrqa/MRQA-Shared-Task-2019", "votes": null }, { "id": "719526", "postDate": "01/15/2020 15:15:09", "content": "https://github.com/huggingface/transformers\nhttps://tfhub.dev/s?module-type=text-embedding&network-architecture=transformer\nhttps://tfhub.dev/google/universal-sentence-encoder/4\nhttps://tfhub.dev/google/universal-sentence-encoder-large/5\nhttp://www.msmarco.org/dataset.aspx\nhttps://github.com/google-research-datasets/boolean-questions\nhttps://github.com/nyu-mll/GLUE-baselines\nhttps://github.com/nyu-mll/jiant/\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttp://www.cs.cmu.edu/~glai1/data/race/\nhttps://tfhub.dev/google/albert_base/3\nhttps://tfhub.dev/google/albert_large/3\nhttps://tfhub.dev/google/albert_xlarge/3\nhttps://tfhub.dev/google/albert_xxlarge/3\nhttps://github.com/zihangdai/xlnet\nhttps://www.tensorflow.org/datasets/catalog/overview\nhttps://ai.google.com/research/NaturalQuestions/download
", "rawMarkdown": "https://github.com/huggingface/transformers\nhttps://tfhub.dev/s?module-type=text-embedding&network-architecture=transformer\nhttps://tfhub.dev/google/universal-sentence-encoder/4\nhttps://tfhub.dev/google/universal-sentence-encoder-large/5\nhttp://www.msmarco.org/dataset.aspx\nhttps://github.com/google-research-datasets/boolean-questions\nhttps://github.com/nyu-mll/GLUE-baselines\nhttps://github.com/nyu-mll/jiant/\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttp://www.cs.cmu.edu/~glai1/data/race/\nhttps://tfhub.dev/google/albert_base/3\nhttps://tfhub.dev/google/albert_large/3\nhttps://tfhub.dev/google/albert_xlarge/3\nhttps://tfhub.dev/google/albert_xxlarge/3\nhttps://github.com/zihangdai/xlnet\nhttps://www.tensorflow.org/datasets/catalog/overview\nhttps://ai.google.com/research/NaturalQuestions/download", "votes": null }, { "id": "719586", "postDate": "01/15/2020 16:13:48", "content": "https://github.com/google-research-datasets/natural-questions\nhttps://github.com/google-research/language/tree/master/language/question_answering/bert_joint\nhttps://rajpurkar.github.io/SQuAD-explorer/\nhttps://github.com/huggingface/transformers\nhttps://huggingface.co/models
", "rawMarkdown": "https://github.com/google-research-datasets/natural-questions\nhttps://github.com/google-research/language/tree/master/language/question_answering/bert_joint\nhttps://rajpurkar.github.io/SQuAD-explorer/\nhttps://github.com/huggingface/transformers\nhttps://huggingface.co/models", "votes": null }, { "id": "719798", "postDate": "01/15/2020 21:28:57", "content": "I'm currently using or may use some of the following external datasets and pretrained models:
\n\ndatasets:
https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs
pretrained models:
https://github.com/google-research/language/tree/master/language/question_answering
I use the original NQ dataset (dev)
\n\nhttps://ai.google.com/research/NaturalQuestions
\n\nOthers are from Hugging Face's transformers package
\n\nhttps://github.com/huggingface/transformers
\n\nAll files can be found in
\n\nhttps://www.kaggle.com/yihdarshieh/nq-competition/metadata
", "rawMarkdown": "I use the original NQ dataset (dev)\n\n[https://ai.google.com/research/NaturalQuestions](https://ai.google.com/research/NaturalQuestions)\n\nOthers are from Hugging Face's transformers package\n\n[https://github.com/huggingface/transformers](https://github.com/huggingface/transformers)\n\nAll files can be found in\n\n[https://www.kaggle.com/yihdarshieh/nq-competition/metadata](https://www.kaggle.com/yihdarshieh/nq-competition/metadata)", "votes": null }, { "id": "721080", "postDate": "01/17/2020 02:53:55", "content": "Original NQ dataset from https://ai.google.com/research/NaturalQuestions v
\n\nhttps://github.com/huggingface/transformers
", "rawMarkdown": "Original NQ dataset from https://ai.google.com/research/NaturalQuestions v\n\nhttps://github.com/huggingface/transformers", "votes": null }, { "id": "725916", "postDate": "01/22/2020 15:59:06", "content": "pretrained models\nhttps://github.com/google-research/language/tree/master/language/question_answering
", "rawMarkdown": "pretrained models\nhttps://github.com/google-research/language/tree/master/language/question_answering", "votes": null }, { "id": "1063856", "postDate": "10/29/2020 12:22:40", "content": "Hi, I am little new to BERT, can someone point me to an example where input data is in csv format - just question and all possible answer from then , test data which will have only question and answers which will be very similar to one or more answers in train data. and we would like to predict the best correct answer for each of the questions in test data.
", "rawMarkdown": "Hi, I am little new to BERT, can someone point me to an example where input data is in csv format - just question and all possible answer from then , test data which will have only question and answers which will be very similar to one or more answers in train data. and we would like to predict the best correct answer for each of the questions in test data.", "votes": null } ], "comments": [ { "id": 1063856, "author_name": "ajeeto", "author_url": "", "post_date": "10/29/2020 12:22:40", "content": "Hi, I am little new to BERT, can someone point me to an example where input data is in csv format - just question and all possible answer from then , test data which will have only question and answers which will be very similar to one or more answers in train data. and we would like to predict the best correct answer for each of the questions in test data.
", "votes": null, "replies": [] }, { "id": 664080, "author_name": "atulkum", "author_url": "", "post_date": "11/03/2019 05:55:57", "content": "Do we need to use tensorflow 2.0 or we are free to use any other framework?\nAlso how we can upload the trained model?
", "votes": null, "replies": [ { "id": 665122, "author_name": "juliaelliott", "author_url": "", "post_date": "11/04/2019 16:46:58", "content": "You are free to use any framework you choose. There are, however, a special set of prizes specific to use of TF2.0, which you can only be eligible for if you meet those requirements.
", "votes": null, "replies": [] } ] }, { "id": 675311, "author_name": "alestainer", "author_url": "", "post_date": "11/17/2019 22:19:32", "content": "Do you consider pretrained models to be external data?
", "votes": null, "replies": [] }, { "id": 681202, "author_name": "hiteiteki", "author_url": "", "post_date": "11/25/2019 20:32:02", "content": "Hello,\nI decided to enter this competition to freshen up my Python and ML skills. I thought live streaming my programming/modelling sessions would provide additional motivation to follow through to that goal. So as per public code sharing rules, I have to share the live streaming session link:
\n\n\n\nThis will be a semi-regular stream, but I plan on streaming the entire effort.
\n\nDisclaimer: due to the emphemeral nature of live video streaming, the videos may or may not disappear in the future. If you wish to save the broadcast, I suggest to use software made for that purpose.
", "votes": null, "replies": [] }, { "id": 685504, "author_name": "dimitreoliveira", "author_url": "", "post_date": "12/01/2019 19:54:17", "content": "What's the \"No custom packages enabled in kernels\" means ? Can I use \"sys.path.append(xxx)\" ?
", "votes": null, "replies": [] }, { "id": 696143, "author_name": "httpwwwfszyc", "author_url": "", "post_date": "12/16/2019 07:13:20", "content": "one question: is the 3h limit of GPU counted only for the 88% private data or for the 100% whole data (public + private test data) ?
", "votes": null, "replies": [] }, { "id": 703214, "author_name": "ddanevskyi", "author_url": "", "post_date": "12/25/2019 19:52:23", "content": "https://github.com/huggingface/transformers
", "votes": null, "replies": [] }, { "id": 706470, "author_name": "irustandi", "author_url": "", "post_date": "12/30/2019 12:43:05", "content": "All the models in https://github.com/huggingface/transformers
", "votes": null, "replies": [] }, { "id": 709925, "author_name": "ejmejm", "author_url": "", "post_date": "01/04/2020 04:50:21", "content": "https://github.com/kamalkraj/ALBERT-TF2.0
", "votes": null, "replies": [] }, { "id": 711498, "author_name": "tonyxu", "author_url": "", "post_date": "01/06/2020 07:31:38", "content": "https://nlp.cs.washington.edu/triviaqa/
", "votes": null, "replies": [] }, { "id": 714594, "author_name": "kentaronakanishi", "author_url": "", "post_date": "01/09/2020 15:25:51", "content": "SQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers
", "votes": null, "replies": [] }, { "id": 715985, "author_name": "httpwwwfszyc", "author_url": "", "post_date": "01/11/2020 05:08:30", "content": "Used or might be used:\nSQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers\nhttps://github.com/bojone/bert4keras\nhttps://github.com/CyberZHG/keras-bert\nhttps://ai.google.com/research/NaturalQuestions\nhttps://github.com/google-research/language\nand other datasets announced by other competitors of this match.
", "votes": null, "replies": [] }, { "id": 716015, "author_name": "snakayama", "author_url": "", "post_date": "01/11/2020 06:33:18", "content": "https://github.com/ThilinaRajapakse/simpletransformers
", "votes": null, "replies": [] }, { "id": 717902, "author_name": "josefslavicek", "author_url": "", "post_date": "01/13/2020 18:57:10", "content": "https://www.kaggle.com/prokaj/bert-joint-baseline
\n\nhttps://www.kaggle.com/mmmarchetti/bert-joint
\n\ncopy of some python files from pip install sentencepiece bert-for-tf2
https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/1
\n\nmaybe I will try something from:
\n\n\n\n\n\nhttps://github.com/google-research/bert
\n\nhttps://github.com/google-research/ALBERT
\n\nhttps://github.com/nshepperd/gpt-2
", "votes": null, "replies": [] }, { "id": 719024, "author_name": "rishigami", "author_url": "", "post_date": "01/15/2020 03:22:43", "content": "I may or may not use these models / external datasets ヽ( ˊᵕˋ )ノ\nhttps://github.com/pytorch/fairseq/tree/master/examples/roberta\nhttps://github.com/pytorch/fairseq/tree/master/examples/bart\nhttps://github.com/zihangdai/xlnet\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttps://github.com/PaddlePaddle/ERNIE\nhttps://github.com/NervanaSystems/nlp-architect\nhttps://github.com/huawei-noah/Pretrained-Language-Model/tree/master/TinyBERT\nhttps://github.com/cooelf/SemBERT\nhttps://github.com/mandarjoshi90/coref\nhttps://github.com/facebookresearch/SpanBERT\nhttps://github.com/alexa/wqa_tanda\nhttps://github.com/mrqa/MRQA-Shared-Task-2019
", "votes": null, "replies": [] }, { "id": 719526, "author_name": "chewkokwahibrainai", "author_url": "", "post_date": "01/15/2020 15:15:09", "content": "https://github.com/huggingface/transformers\nhttps://tfhub.dev/s?module-type=text-embedding&network-architecture=transformer\nhttps://tfhub.dev/google/universal-sentence-encoder/4\nhttps://tfhub.dev/google/universal-sentence-encoder-large/5\nhttp://www.msmarco.org/dataset.aspx\nhttps://github.com/google-research-datasets/boolean-questions\nhttps://github.com/nyu-mll/GLUE-baselines\nhttps://github.com/nyu-mll/jiant/\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttp://www.cs.cmu.edu/~glai1/data/race/\nhttps://tfhub.dev/google/albert_base/3\nhttps://tfhub.dev/google/albert_large/3\nhttps://tfhub.dev/google/albert_xlarge/3\nhttps://tfhub.dev/google/albert_xxlarge/3\nhttps://github.com/zihangdai/xlnet\nhttps://www.tensorflow.org/datasets/catalog/overview\nhttps://ai.google.com/research/NaturalQuestions/download
", "votes": null, "replies": [] }, { "id": 719586, "author_name": "noxuslol", "author_url": "", "post_date": "01/15/2020 16:13:48", "content": "https://github.com/google-research-datasets/natural-questions\nhttps://github.com/google-research/language/tree/master/language/question_answering/bert_joint\nhttps://rajpurkar.github.io/SQuAD-explorer/\nhttps://github.com/huggingface/transformers\nhttps://huggingface.co/models
", "votes": null, "replies": [] }, { "id": 719798, "author_name": "aagapi", "author_url": "", "post_date": "01/15/2020 21:28:57", "content": "I'm currently using or may use some of the following external datasets and pretrained models:
\n\ndatasets:
https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs
pretrained models:
https://github.com/google-research/language/tree/master/language/question_answering
I use the original NQ dataset (dev)
\n\nhttps://ai.google.com/research/NaturalQuestions
\n\nOthers are from Hugging Face's transformers package
\n\nhttps://github.com/huggingface/transformers
\n\nAll files can be found in
\n\nhttps://www.kaggle.com/yihdarshieh/nq-competition/metadata
", "votes": null, "replies": [] }, { "id": 721080, "author_name": "wochidadonggua", "author_url": "", "post_date": "01/17/2020 02:53:55", "content": "Original NQ dataset from https://ai.google.com/research/NaturalQuestions v
\n\nhttps://github.com/huggingface/transformers
", "votes": null, "replies": [] }, { "id": 725916, "author_name": "kcotton21", "author_url": "", "post_date": "01/22/2020 15:59:06", "content": "pretrained models\nhttps://github.com/google-research/language/tree/master/language/question_answering
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "660248": "Per the [Competition Rules](https://www.kaggle.com/c/tensorflow2-question-answering/rules), freely and publicly available external data is permitted in this competition, but must be posted to this forum thread no later than the Entry Deadline (one week before competition close).\n\nOnce someone posts an external dataset to this thread, you do not need to re-post it if you are using the same one.", "664080": "Do we need to use tensorflow 2.0 or we are free to use any other framework?\nAlso how we can upload the trained model?", "665122": "You are free to use any framework you choose. There are, however, a [special set of prizes specific to use of TF2.0](https://www.kaggle.com/c/tensorflow2-question-answering/overview/prizes), which you can only be eligible for if you meet those requirements.", "675311": "Do you consider pretrained models to be external data?", "681202": "Hello,\nI decided to enter this competition to freshen up my Python and ML skills. I thought live streaming my programming/modelling sessions would provide additional motivation to follow through to that goal. So as per public code sharing rules, I have to share the live streaming session link:\n\nhttps://www.twitch.tv/cythie\n\nThis will be a semi-regular stream, but I plan on streaming the entire effort.\n\nDisclaimer: due to the emphemeral nature of live video streaming, the videos may or may not disappear in the future. If you wish to save the broadcast, I suggest to use software made for that purpose.", "685504": "https://github.com/google-research/language/tree/master/language/question_answering/bert_joint\n- https://github.com/tensorflow/models/tree/master/official/nlp", "692624": "What's the \"No custom packages enabled in kernels\" means ? Can I use \"sys.path.append(xxx)\" ?", "696143": "one question: is the 3h limit of GPU counted only for the 88% private data or for the 100% whole data (public + private test data) ?", "703214": "https://github.com/huggingface/transformers", "706470": "All the models in https://github.com/huggingface/transformers", "709925": "https://github.com/kamalkraj/ALBERT-TF2.0", "711498": "https://github.com/google-research/language/tree/master/language/question_answering/bert_joint\n- https://github.com/google-research/bert\n- https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json\n- https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v1.1.json", "714107": "https://nlp.cs.washington.edu/triviaqa/", "714594": "SQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers", "715985": "Used or might be used:\nSQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/\nhuggingface transformers https://github.com/huggingface/transformers\nhttps://github.com/bojone/bert4keras\nhttps://github.com/CyberZHG/keras-bert\nhttps://ai.google.com/research/NaturalQuestions\nhttps://github.com/google-research/language\nand other datasets announced by other competitors of this match.", "716015": "https://github.com/ThilinaRajapakse/simpletransformers", "717902": "https://www.kaggle.com/prokaj/bert-joint-baseline\n\nhttps://www.kaggle.com/mmmarchetti/bert-joint\n\ncopy of some python files from ```pip install sentencepiece bert-for-tf2```\n\nhttps://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/1\n\nmaybe I will try something from:\n\nhttps://tfhub.dev/s?q=bert\n\nhttps://tfhub.dev/s?q=albert\n\nhttps://github.com/google-research/bert\n\nhttps://github.com/google-research/ALBERT\n\nhttps://github.com/nshepperd/gpt-2", "719024": "I may or may not use these models / external datasets ヽ( ˊᵕˋ )ノ\nhttps://github.com/pytorch/fairseq/tree/master/examples/roberta\nhttps://github.com/pytorch/fairseq/tree/master/examples/bart\nhttps://github.com/zihangdai/xlnet\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttps://github.com/PaddlePaddle/ERNIE\nhttps://github.com/NervanaSystems/nlp-architect\nhttps://github.com/huawei-noah/Pretrained-Language-Model/tree/master/TinyBERT\nhttps://github.com/cooelf/SemBERT\nhttps://github.com/mandarjoshi90/coref\nhttps://github.com/facebookresearch/SpanBERT\nhttps://github.com/alexa/wqa_tanda\nhttps://github.com/mrqa/MRQA-Shared-Task-2019", "719526": "https://github.com/huggingface/transformers\nhttps://tfhub.dev/s?module-type=text-embedding&network-architecture=transformer\nhttps://tfhub.dev/google/universal-sentence-encoder/4\nhttps://tfhub.dev/google/universal-sentence-encoder-large/5\nhttp://www.msmarco.org/dataset.aspx\nhttps://github.com/google-research-datasets/boolean-questions\nhttps://github.com/nyu-mll/GLUE-baselines\nhttps://github.com/nyu-mll/jiant/\nhttps://github.com/google-research/text-to-text-transfer-transformer\nhttp://www.cs.cmu.edu/~glai1/data/race/\nhttps://tfhub.dev/google/albert_base/3\nhttps://tfhub.dev/google/albert_large/3\nhttps://tfhub.dev/google/albert_xlarge/3\nhttps://tfhub.dev/google/albert_xxlarge/3\nhttps://github.com/zihangdai/xlnet\nhttps://www.tensorflow.org/datasets/catalog/overview\nhttps://ai.google.com/research/NaturalQuestions/download", "719586": "https://github.com/google-research-datasets/natural-questions\nhttps://github.com/google-research/language/tree/master/language/question_answering/bert_joint\nhttps://rajpurkar.github.io/SQuAD-explorer/\nhttps://github.com/huggingface/transformers\nhttps://huggingface.co/models", "719798": "I'm currently using or may use some of the following external datasets and pretrained models:\n\n- datasets:\n\n- https://rajpurkar.github.io/SQuAD-explorer\n- https://gluebenchmark.com/\n- https://nlp.stanford.edu/projects/snli/\n- https://www.nyu.edu/projects/bowman/xnli/\n- http://www.nyu.edu/projects/bowman/multinli/\n- https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs\n\n- pretrained models:\n\n- https://github.com/google-research/language/tree/master/language/question_answering\n- https://github.com/tensorflow/models/tree/master/official/nlp\n- https://github.com/huggingface/transformers\n- https://github.com/google-research/bert\n- https://github.com/pytorch/fairseq/tree/master/examples/roberta\n- https://github.com/kamalkraj/ALBERT-TF2.0\n- https://tfhub.dev/s?module-type=text-embedding", "720349": "https://dumps.wikimedia.org/", "721013": "I use the original NQ dataset (dev)\n\n[https://ai.google.com/research/NaturalQuestions](https://ai.google.com/research/NaturalQuestions)\n\nOthers are from Hugging Face's transformers package\n\n[https://github.com/huggingface/transformers](https://github.com/huggingface/transformers)\n\nAll files can be found in\n\n[https://www.kaggle.com/yihdarshieh/nq-competition/metadata](https://www.kaggle.com/yihdarshieh/nq-competition/metadata)", "721080": "Original NQ dataset from https://ai.google.com/research/NaturalQuestions v\n\nhttps://github.com/huggingface/transformers", "725916": "pretrained models\nhttps://github.com/google-research/language/tree/master/language/question_answering", "1063856": "Hi, I am little new to BERT, can someone point me to an example where input data is in csv format - just question and all possible answer from then , test data which will have only question and answers which will be very similar to one or more answers in train data. and we would like to predict the best correct answer for each of the questions in test data." }, "source": "meta" }