![\"Example\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2591653%2F1e697fb487e8d84457a9da37550d1587%2FScreenshot%202020-11-30%20225739.png?generation=1606757279804982&alt=media\")
Since in this competition, we need to make the inference part without using the internet, we are not able to install packages using the usual pip command like:
To install EfficientNet package:
\n!pip install --quiet efficientnet
But we can still install packages like EfficientNet using the dataset feature from Kaggle you can see a working example at this inference notebook.
EfficientNetpip with !pip install --quiet path_to_package!pip install --quiet /kaggle/input/efficientnet-gitimport efficientnet.tfkeras as efnSince in this competition, we need to make the inference part without using the internet, we are not able to install packages using the usual pip command like:
To install EfficientNet package:
\n!pip install --quiet efficientnet
But we can still install packages like EfficientNet using the dataset feature from Kaggle you can see a working example at this inference notebook.
EfficientNetpip with !pip install --quiet path_to_package!pip install --quiet /kaggle/input/efficientnet-gitimport efficientnet.tfkeras as efnVery rare and helpful notebook, learn new things, thanks for sharing😃.
", "rawMarkdown": "Very rare and helpful notebook, learn new things, thanks for sharing😃.", "votes": 1 }, { "id": 1119244, "postDate": "2020-12-19T20:53:56.160Z", "content": "Great, didn't know that!
", "rawMarkdown": "Great, didn't know that!", "votes": 1 }, { "id": 1111332, "postDate": "2020-12-13T16:25:33.770Z", "content": "Somehow not working for me. Gives me error:
\nERROR: No matching distribution found for keras_applications=1.0.7 (from efficientnet==1.1.1)
\nAny ideas why?
Hey @serg132003 take a look at my notebook here it is working for me, have you added the datasets with the repositories to the notebook?
", "rawMarkdown": "Hey @serg132003 take a look at my notebook [here](https://www.kaggle.com/dimitreoliveira/cassava-leaf-disease-tpu-tensorflow-inference/notebook) it is working for me, have you added the datasets with the repositories to the notebook?" } ] }, { "id": 1096676, "postDate": "2020-11-30T17:29:48.243Z", "content": "What I am doing is copying the required whl files from pip into a dataset and doing
\n!pip install [PATH TO WHEEL] at the start of the notebook as shown in the image below
Nice @abhinand05 , this is very close to a solution, but in my case, the pip command installs the package using the setup.py file.
I'm still not able to use the EfficientNet model. These are the steps I followed.
\n! pip3 download efficientnet_pytorch which downloads the following files to /kaggle/working folder
numpy-1.19.4-cp37-cp37m-manylinux2010_x86_64.whl\ntorch-1.7.0-cp37-cp37m-manylinux1_x86_64.whl\nefficientnet_pytorch-0.7.0.tar.gz\ndataclasses-0.6-py3-none-any.whl\nfuture-0.18.2.tar.gz\ntyping_extensions-3.7.4.3-py3-none-any.whl\nDownloaded the 4 .whl files to my pc. and Uploaded them to dataset and I get the warnings for all 4 whl file which I have attached in the screenshot. Please help
", "rawMarkdown": "@abhinand05\n\nI'm still not able to use the EfficientNet model. These are the steps I followed. \n```! pip3 download efficientnet_pytorch``` which downloads the following files to ```/kaggle/working``` folder\n\n```\nnumpy-1.19.4-cp37-cp37m-manylinux2010_x86_64.whl\ntorch-1.7.0-cp37-cp37m-manylinux1_x86_64.whl\nefficientnet_pytorch-0.7.0.tar.gz\ndataclasses-0.6-py3-none-any.whl\nfuture-0.18.2.tar.gz\ntyping_extensions-3.7.4.3-py3-none-any.whl\n```\n\nDownloaded the 4 .whl files to my pc. and Uploaded them to dataset and I get the warnings for all 4 whl file which I have attached in the screenshot. Please help\n\n\n\n" } ] }, { "id": 1095816, "postDate": "2020-11-30T01:28:31.350Z", "content": "A good tip! Another option is to search for the actual source code for the models you wish to use, and simply paste it into your notebook.
", "rawMarkdown": "A good tip! Another option is to search for the actual source code for the models you wish to use, and simply paste it into your notebook.", "votes": 1, "replies": [ { "id": 1095832, "postDate": "2020-11-30T01:54:37.123Z", "content": "Yes @thomasbrekkunnvik , this works too, I usually do that for small functions, but in the case of large packages like EfficientNet, I think the code would get too messy.
", "rawMarkdown": "Yes @thomasbrekkunnvik , this works too, I usually do that for small functions, but in the case of large packages like EfficientNet, I think the code would get too messy.", "votes": 1 } ] }, { "id": 1088679, "postDate": "2020-11-23T21:10:11.513Z", "content": "EfficientNet is using the only one file (imagenet_utils.py) from keras-applications package, so one can also copy this file into EfficientNet's folder.
", "rawMarkdown": "EfficientNet is using the only one file (imagenet_utils.py) from keras-applications package, so one can also copy this file into EfficientNet's folder.", "votes": 2, "replies": [ { "id": 1095830, "postDate": "2020-11-30T01:53:36.650Z", "content": "That is a good point @sergeyzlobin , I had not noted that.
", "rawMarkdown": "That is a good point @sergeyzlobin , I had not noted that." } ] }, { "id": 1098262, "postDate": "2020-12-01T14:28:29.867Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1108166, "postDate": "2020-12-10T10:55:51.417Z", "content": "Thanks! Solved my day!
", "rawMarkdown": "Thanks! Solved my day!", "votes": 1 }, { "id": 1099754, "postDate": "2020-12-02T15:28:20.623Z", "content": "Thanks a lot. It is really helpful.
", "rawMarkdown": "Thanks a lot. It is really helpful. ", "votes": 1 }, { "id": 1096399, "postDate": "2020-11-30T13:23:14.773Z", "content": "Thanks, it helps me a lot :D
", "rawMarkdown": "Thanks, it helps me a lot :D", "votes": 1 }, { "id": 1095634, "postDate": "2020-11-29T19:49:02.967Z", "content": "Thanks, it works!
", "rawMarkdown": "Thanks, it works!", "votes": 1 }, { "id": 1088363, "postDate": "2020-11-23T14:59:57.110Z", "content": "Thanks for sharing!
", "rawMarkdown": "Thanks for sharing!", "votes": 1 } ], "comments": [ { "id": 1121497, "author_name": "Sudhanshu Singh", "author_url": "", "post_date": "2020-12-21T17:41:13.220000", "content": "Very rare and helpful notebook, learn new things, thanks for sharing😃.
", "votes": 1, "replies": [] }, { "id": 1119244, "author_name": "Bouwe Ceunen", "author_url": "", "post_date": "2020-12-19T20:53:56.160000", "content": "Great, didn't know that!
", "votes": 1, "replies": [] }, { "id": 1111332, "author_name": "Serge", "author_url": "", "post_date": "2020-12-13T16:25:33.770000", "content": "Somehow not working for me. Gives me error:
\nERROR: No matching distribution found for keras_applications=1.0.7 (from efficientnet==1.1.1)
\nAny ideas why?
Hey @serg132003 take a look at my notebook here it is working for me, have you added the datasets with the repositories to the notebook?
", "votes": 0, "replies": [] } ] }, { "id": 1096676, "author_name": "Abhinand", "author_url": "", "post_date": "2020-11-30T17:29:48.243000", "content": "What I am doing is copying the required whl files from pip into a dataset and doing
\n!pip install [PATH TO WHEEL] at the start of the notebook as shown in the image below
Nice @abhinand05 , this is very close to a solution, but in my case, the pip command installs the package using the setup.py file.
I'm still not able to use the EfficientNet model. These are the steps I followed.
\n! pip3 download efficientnet_pytorch which downloads the following files to /kaggle/working folder
numpy-1.19.4-cp37-cp37m-manylinux2010_x86_64.whl\ntorch-1.7.0-cp37-cp37m-manylinux1_x86_64.whl\nefficientnet_pytorch-0.7.0.tar.gz\ndataclasses-0.6-py3-none-any.whl\nfuture-0.18.2.tar.gz\ntyping_extensions-3.7.4.3-py3-none-any.whl\nDownloaded the 4 .whl files to my pc. and Uploaded them to dataset and I get the warnings for all 4 whl file which I have attached in the screenshot. Please help
", "votes": 0, "replies": [] } ] }, { "id": 1095816, "author_name": "Thomas Brekk Unnvik", "author_url": "", "post_date": "2020-11-30T01:28:31.350000", "content": "A good tip! Another option is to search for the actual source code for the models you wish to use, and simply paste it into your notebook.
", "votes": 1, "replies": [ { "id": 1095832, "author_name": "DimitreOliveira", "author_url": "", "post_date": "2020-11-30T01:54:37.123000", "content": "Yes @thomasbrekkunnvik , this works too, I usually do that for small functions, but in the case of large packages like EfficientNet, I think the code would get too messy.
", "votes": 1, "replies": [] } ] }, { "id": 1088679, "author_name": "Sergey Zlobin", "author_url": "", "post_date": "2020-11-23T21:10:11.513000", "content": "EfficientNet is using the only one file (imagenet_utils.py) from keras-applications package, so one can also copy this file into EfficientNet's folder.
", "votes": 2, "replies": [ { "id": 1095830, "author_name": "DimitreOliveira", "author_url": "", "post_date": "2020-11-30T01:53:36.650000", "content": "That is a good point @sergeyzlobin , I had not noted that.
", "votes": 0, "replies": [] } ] }, { "id": 1098262, "author_name": "", "author_url": "", "post_date": "2020-12-01T14:28:29.867000", "content": "", "votes": 0, "replies": [] }, { "id": 1108166, "author_name": "Giovanni Pinamonti", "author_url": "", "post_date": "2020-12-10T10:55:51.417000", "content": "Thanks! Solved my day!
", "votes": 1, "replies": [] }, { "id": 1099754, "author_name": "Deepak Rai", "author_url": "", "post_date": "2020-12-02T15:28:20.623000", "content": "Thanks a lot. It is really helpful.
", "votes": 1, "replies": [] }, { "id": 1096399, "author_name": "Rakka Alhazimi", "author_url": "", "post_date": "2020-11-30T13:23:14.773000", "content": "Thanks, it helps me a lot :D
", "votes": 1, "replies": [] }, { "id": 1095634, "author_name": "Anton Morenov", "author_url": "", "post_date": "2020-11-29T19:49:02.967000", "content": "Thanks, it works!
", "votes": 1, "replies": [] }, { "id": 1088363, "author_name": "Mau Rua", "author_url": "", "post_date": "2020-11-23T14:59:57.110000", "content": "Thanks for sharing!
", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "1087527": "Since in this competition, we need to make the inference part without using the internet, we are not able to install packages using the usual `pip` command like:\n\nTo install [EfficientNet](https://github.com/qubvel/efficientnet) package:\n`!pip install --quiet efficientnet`\n\nBut we can still install packages like `EfficientNet` using the dataset feature from Kaggle you can see a working example at this [inference notebook](https://www.kaggle.com/dimitreoliveira/cassava-leaf-disease-tpu-tensorflow-inference).\n\n#### Steps:\n1. Create a dataset from the GitGub repository, I have created the two that you will need to use `EfficientNet`\n - [keras-applications](https://www.kaggle.com/dimitreoliveira/kerasapplications) package.\n - [efficientNet](https://www.kaggle.com/dimitreoliveira/efficientnet-git) package.\n2. Install the package using `pip` with `!pip install --quiet path_to_package`\n - Example: `!pip install --quiet /kaggle/input/efficientnet-git`\n3. Import the package\n - Example `import efficientnet.tfkeras as efn`", "1121497": "Very rare and helpful notebook, learn new things, thanks for sharing😃.", "1119244": "Great, didn't know that!", "1111332": "Somehow not working for me. Gives me error:\nERROR: No matching distribution found for keras_applications<=1.0.8,>=1.0.7 (from efficientnet==1.1.1)\nAny ideas why?", "1096676": "What I am doing is copying the required `whl` files from pip into a dataset and doing \n`!pip install [PATH TO WHEEL]` at the start of the notebook as shown in the image below\n\n\n\n", "1095816": "A good tip! Another option is to search for the actual source code for the models you wish to use, and simply paste it into your notebook.", "1088679": "EfficientNet is using the only one file (imagenet_utils.py) from keras-applications package, so one can also copy this file into EfficientNet's folder.", "1098262": "", "1108166": "Thanks! Solved my day!", "1099754": "Thanks a lot. It is really helpful. ", "1096399": "Thanks, it helps me a lot :D", "1095634": "Thanks, it works!", "1088363": "Thanks for sharing!" } }