The organizers say :
\n\n\n\n\nYour model must be named submission.zip and be compatible with TensorFlow 2.2.
\n
Even if tensorflow is a great tool and spread among the Kaggle community, I think It would be better to open this competiton to other Kagglers that are not used to tensorflow ie those from pytorch, caffe, mxnet ...
\n\nThis more obvious again when we recall that there is the ONNX format for neural net exchange between different frameworks ...
", "messages": [ { "id": 910121, "postDate": "2020-07-01T02:17:17.487Z", "content": "The organizers say :
\n\n\n\n\nYour model must be named submission.zip and be compatible with TensorFlow 2.2.
\n
Even if tensorflow is a great tool and spread among the Kaggle community, I think It would be better to open this competiton to other Kagglers that are not used to tensorflow ie those from pytorch, caffe, mxnet ...
\n\nThis more obvious again when we recall that there is the ONNX format for neural net exchange between different frameworks ...
", "rawMarkdown": "The organizers say : \n\n> Your model must be named submission.zip and be compatible with TensorFlow 2.2. \n\nEven if **tensorflow** is a great tool and spread among the Kaggle community, I think It would be better to open this competiton to other Kagglers that are not used to **tensorflow** ie those from **pytorch**, **caffe**, **mxnet** ... \n\nThis more obvious again when we recall that there is the **ONNX** format for neural net exchange between different frameworks ...", "votes": 21 }, { "id": 910137, "postDate": "2020-07-01T02:28:17.937Z", "content": "Google Landmark Recognition, after all. Emphasis on GOOGLE.
", "rawMarkdown": "Google Landmark Recognition, after all. Emphasis on **GOOGLE.**", "votes": 7, "replies": [ { "id": 910378, "postDate": "2020-07-01T05:59:31.040Z", "rawMarkdown": "", "isDeleted": true }, { "id": 910437, "postDate": "2020-07-01T06:38:25.427Z", "content": "There is no update over there https://www.kaggle.com/c/landmark-retrieval-2020/data .
", "rawMarkdown": "There is no update over there https://www.kaggle.com/c/landmark-retrieval-2020/data .", "votes": 1 }, { "id": 910461, "postDate": "2020-07-01T06:50:30.143Z", "content": "Ah, my bad. I didn't knew this was stated in the data description. Sorry for the inconvenience caused.
", "rawMarkdown": "Ah, my bad. I didn't knew this was stated in the data description. Sorry for the inconvenience caused." }, { "id": 911419, "postDate": "2020-07-01T18:05:46.507Z", "content": "We had a Facebook Deep Fake Detection Competition here and there was no obligation to use pytorch !
", "rawMarkdown": "We had a Facebook Deep Fake Detection Competition here and there was no obligation to use pytorch !", "votes": 1 }, { "id": 911437, "postDate": "2020-07-01T18:25:48.053Z", "content": "It has been confirmed that there is no restriction on training with TF 2.2, you can train with PyTorch and convert to a TensorFlow-compatible model using ONNX. :)
\n\nHere is the thread: https://www.kaggle.com/c/landmark-retrieval-2020/discussion/163139
", "rawMarkdown": "It has been confirmed that there is no restriction on training with TF 2.2, you can train with PyTorch and convert to a TensorFlow-compatible model using ONNX. :)\n\nHere is the thread: https://www.kaggle.com/c/landmark-retrieval-2020/discussion/163139", "votes": 1 } ] }, { "id": 911126, "postDate": "2020-07-01T15:26:39.347Z", "content": "Sorry, I can't find the thread where this is mentioned, could you please point me to it?\nThanks!
\n\nEdit: Found it, Sorry: https://www.kaggle.com/c/landmark-retrieval-2020/data
\n\nI was looking at rules instead.
", "rawMarkdown": "Sorry, I can't find the thread where this is mentioned, could you please point me to it?\nThanks!\n\nEdit: Found it, Sorry: https://www.kaggle.com/c/landmark-retrieval-2020/data\n\nI was looking at rules instead.", "votes": 1 }, { "id": 911375, "postDate": "2020-07-01T17:32:36.503Z", "content": "This is because its Google after all and they want to promote their Tensorflow(specially the newly released 2.X version) - [Completely my opinion]
\n\nWhat I think is Google wants users to learn/focus on the changes there are in this new 2.X version and how its implementations are better than pytorch, caffe, etc. But at the same time this will make the competition less attractive and have less people participating :(
", "rawMarkdown": "This is because its Google after all and they want to promote their Tensorflow(specially the newly released 2.X version) - [Completely my opinion]\n\nWhat I think is Google wants users to learn/focus on the changes there are in this new 2.X version and how its implementations are better than pytorch, caffe, etc. But at the same time this will make the competition less attractive and have less people participating :(\n", "votes": 2 }, { "id": 921897, "postDate": "2020-07-09T16:28:34.943Z", "content": "Why is such a requirement there?
", "rawMarkdown": "Why is such a requirement there?" }, { "id": 911890, "postDate": "2020-07-02T04:58:41.940Z", "content": "I am new To kaggle please Guide Me
", "rawMarkdown": "I am new To kaggle please Guide Me" }, { "id": 910191, "postDate": "2020-07-01T03:33:36.757Z", "content": "I also confuse it,Do I have to train with tensorflow？
", "rawMarkdown": "I also confuse it,Do I have to train with tensorflow？", "replies": [ { "id": 910634, "postDate": "2020-07-01T09:10:56.440Z", "rawMarkdown": "", "isDeleted": true }, { "id": 911711, "postDate": "2020-07-02T00:58:48.283Z", "content": "Thanks for your reply! Mr.White
", "rawMarkdown": "Thanks for your reply! Mr.White" } ] }, { "id": 910707, "postDate": "2020-07-01T10:14:44.560Z", "rawMarkdown": "", "votes": 1, "isDeleted": true, "replies": [ { "id": 911086, "postDate": "2020-07-01T15:02:43.820Z", "content": "Definitely this rectriction will make the competition less attractive :worried:
", "rawMarkdown": "Definitely this rectriction will make the competition less attractive :worried:", "votes": 2 } ] }, { "id": 910425, "postDate": "2020-07-01T06:30:52.047Z", "rawMarkdown": "", "votes": -7, "isDeleted": true } ], "comments": [ { "id": 910137, "author_name": "Trigram", "author_url": "", "post_date": "2020-07-01T02:28:17.937000", "content": "Google Landmark Recognition, after all. Emphasis on GOOGLE.
", "votes": 7, "replies": [ { "id": 910378, "author_name": "", "author_url": "", "post_date": "2020-07-01T05:59:31.040000", "content": "", "votes": 0, "replies": [] }, { "id": 910437, "author_name": "kkiller", "author_url": "", "post_date": "2020-07-01T06:38:25.427000", "content": "There is no update over there https://www.kaggle.com/c/landmark-retrieval-2020/data .
", "votes": 1, "replies": [] }, { "id": 910461, "author_name": "Aadhav Vignesh", "author_url": "", "post_date": "2020-07-01T06:50:30.143000", "content": "Ah, my bad. I didn't knew this was stated in the data description. Sorry for the inconvenience caused.
", "votes": 0, "replies": [] }, { "id": 911419, "author_name": "kkiller", "author_url": "", "post_date": "2020-07-01T18:05:46.507000", "content": "We had a Facebook Deep Fake Detection Competition here and there was no obligation to use pytorch !
", "votes": 1, "replies": [] }, { "id": 911437, "author_name": "Aadhav Vignesh", "author_url": "", "post_date": "2020-07-01T18:25:48.053000", "content": "It has been confirmed that there is no restriction on training with TF 2.2, you can train with PyTorch and convert to a TensorFlow-compatible model using ONNX. :)
\n\nHere is the thread: https://www.kaggle.com/c/landmark-retrieval-2020/discussion/163139
", "votes": 1, "replies": [] } ] }, { "id": 911126, "author_name": "Sanyam Bhutani", "author_url": "", "post_date": "2020-07-01T15:26:39.347000", "content": "Sorry, I can't find the thread where this is mentioned, could you please point me to it?\nThanks!
\n\nEdit: Found it, Sorry: https://www.kaggle.com/c/landmark-retrieval-2020/data
\n\nI was looking at rules instead.
", "votes": 1, "replies": [] }, { "id": 911375, "author_name": "Sarthak Rana", "author_url": "", "post_date": "2020-07-01T17:32:36.503000", "content": "This is because its Google after all and they want to promote their Tensorflow(specially the newly released 2.X version) - [Completely my opinion]
\n\nWhat I think is Google wants users to learn/focus on the changes there are in this new 2.X version and how its implementations are better than pytorch, caffe, etc. But at the same time this will make the competition less attractive and have less people participating :(
", "votes": 2, "replies": [] }, { "id": 921897, "author_name": "Kurian Benoy", "author_url": "", "post_date": "2020-07-09T16:28:34.943000", "content": "Why is such a requirement there?
", "votes": 0, "replies": [] }, { "id": 911890, "author_name": "MUKUL CHANDRA", "author_url": "", "post_date": "2020-07-02T04:58:41.940000", "content": "I am new To kaggle please Guide Me
", "votes": 0, "replies": [] }, { "id": 910191, "author_name": "xiaopeng", "author_url": "", "post_date": "2020-07-01T03:33:36.757000", "content": "I also confuse it,Do I have to train with tensorflow？
", "votes": 0, "replies": [ { "id": 910634, "author_name": "", "author_url": "", "post_date": "2020-07-01T09:10:56.440000", "content": "", "votes": 0, "replies": [] }, { "id": 911711, "author_name": "xiaopeng", "author_url": "", "post_date": "2020-07-02T00:58:48.283000", "content": "Thanks for your reply! Mr.White
", "votes": 0, "replies": [] } ] }, { "id": 910707, "author_name": "", "author_url": "", "post_date": "2020-07-01T10:14:44.560000", "content": "", "votes": 1, "replies": [ { "id": 911086, "author_name": "kkiller", "author_url": "", "post_date": "2020-07-01T15:02:43.820000", "content": "Definitely this rectriction will make the competition less attractive :worried:
", "votes": 2, "replies": [] } ] }, { "id": 910425, "author_name": "", "author_url": "", "post_date": "2020-07-01T06:30:52.047000", "content": "", "votes": -7, "replies": [] } ], "raw_markdown_by_id": { "910121": "The organizers say : \n\n> Your model must be named submission.zip and be compatible with TensorFlow 2.2. \n\nEven if **tensorflow** is a great tool and spread among the Kaggle community, I think It would be better to open this competiton to other Kagglers that are not used to **tensorflow** ie those from **pytorch**, **caffe**, **mxnet** ... \n\nThis more obvious again when we recall that there is the **ONNX** format for neural net exchange between different frameworks ...", "910137": "Google Landmark Recognition, after all. Emphasis on **GOOGLE.**", "911126": "Sorry, I can't find the thread where this is mentioned, could you please point me to it?\nThanks!\n\nEdit: Found it, Sorry: https://www.kaggle.com/c/landmark-retrieval-2020/data\n\nI was looking at rules instead.", "911375": "This is because its Google after all and they want to promote their Tensorflow(specially the newly released 2.X version) - [Completely my opinion]\n\nWhat I think is Google wants users to learn/focus on the changes there are in this new 2.X version and how its implementations are better than pytorch, caffe, etc. But at the same time this will make the competition less attractive and have less people participating :(\n", "921897": "Why is such a requirement there?", "911890": "I am new To kaggle please Guide Me", "910191": "I also confuse it,Do I have to train with tensorflow？", "910707": "", "910425": "" } }