{
  "id": 347998,
  "title": "Tensroflow 2.6.4 does not work properly when keyword argument are passed",
  "url": "/competitions/mayo-clinic-strip-ai/discussion/347998",
  "author_name": "Javad Khorramdel",
  "post_date": "2022-08-26T09:34:20.065000",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I'm facing some trouble with Tensorflow .4.6 which is currently installled in Kaggle GPU notebooks. The full detail about the issue can be found in: <a href=\"url\" target=\"_blank\">https://github.com/tensorflow/tensorflow/issues/57449</a>. <br>\nThe short explaination is that I'm writing a tensorflow pipline for reading the images, when I use tf.cast(image, tf.float32), it works properly, but when I use tf.cast(image, dtype=tf.float32) I get this error:<br>\nTypeError: 'dict' object is not callable</p>\n<p>One possible solution is to pass all the keyword argument without using the keyword itself. It works in simple cases like tf.cast command, but there are lots of commands which take lots of keyword argument and most of them are unnecessary for me (I prefer to use the default value for most of them, and only use the argument that I want to change!). </p>\n<p>Another possible solution is to either upgrade or downgrade the tensorflow version. Since it is mentioned in the competition requirements that the Internet should be disabled, is there any way for doing this? </p>",
  "messages": [
    {
      "id": 1914663,
      "postDate": "2022-08-26T09:34:20.067Z",
      "content": "<p>I'm facing some trouble with Tensorflow .4.6 which is currently installled in Kaggle GPU notebooks. The full detail about the issue can be found in: <a href=\"url\" target=\"_blank\">https://github.com/tensorflow/tensorflow/issues/57449</a>. <br>\nThe short explaination is that I'm writing a tensorflow pipline for reading the images, when I use tf.cast(image, tf.float32), it works properly, but when I use tf.cast(image, dtype=tf.float32) I get this error:<br>\nTypeError: 'dict' object is not callable</p>\n<p>One possible solution is to pass all the keyword argument without using the keyword itself. It works in simple cases like tf.cast command, but there are lots of commands which take lots of keyword argument and most of them are unnecessary for me (I prefer to use the default value for most of them, and only use the argument that I want to change!). </p>\n<p>Another possible solution is to either upgrade or downgrade the tensorflow version. Since it is mentioned in the competition requirements that the Internet should be disabled, is there any way for doing this? </p>",
      "rawMarkdown": "I'm facing some trouble with Tensorflow .4.6 which is currently installled in Kaggle GPU notebooks. The full detail about the issue can be found in: [https://github.com/tensorflow/tensorflow/issues/57449](url). \nThe short explaination is that I'm writing a tensorflow pipline for reading the images, when I use tf.cast(image, tf.float32), it works properly, but when I use tf.cast(image, dtype=tf.float32) I get this error:\nTypeError: 'dict' object is not callable\n\nOne possible solution is to pass all the keyword argument without using the keyword itself. It works in simple cases like tf.cast command, but there are lots of commands which take lots of keyword argument and most of them are unnecessary for me (I prefer to use the default value for most of them, and only use the argument that I want to change!). \n\nAnother possible solution is to either upgrade or downgrade the tensorflow version. Since it is mentioned in the competition requirements that the Internet should be disabled, is there any way for doing this? \n",
      "votes": 2
    },
    {
      "id": 1922537,
      "postDate": "2022-09-01T14:53:40.237Z",
      "content": "<p>Are you sure? as far as I know they didn't change this interface.. Is there anything else that might be causing your problem?</p>",
      "rawMarkdown": "Are you sure? as far as I know they didn't change this interface.. Is there anything else that might be causing your problem?"
    }
  ],
  "comments": [
    {
      "id": 1922537,
      "author_name": "The Devastator",
      "author_url": "",
      "post_date": "2022-09-01T14:53:40.237000",
      "content": "<p>Are you sure? as far as I know they didn't change this interface.. Is there anything else that might be causing your problem?</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1914663": "I'm facing some trouble with Tensorflow .4.6 which is currently installled in Kaggle GPU notebooks. The full detail about the issue can be found in: [https://github.com/tensorflow/tensorflow/issues/57449](url). \nThe short explaination is that I'm writing a tensorflow pipline for reading the images, when I use tf.cast(image, tf.float32), it works properly, but when I use tf.cast(image, dtype=tf.float32) I get this error:\nTypeError: 'dict' object is not callable\n\nOne possible solution is to pass all the keyword argument without using the keyword itself. It works in simple cases like tf.cast command, but there are lots of commands which take lots of keyword argument and most of them are unnecessary for me (I prefer to use the default value for most of them, and only use the argument that I want to change!). \n\nAnother possible solution is to either upgrade or downgrade the tensorflow version. Since it is mentioned in the competition requirements that the Internet should be disabled, is there any way for doing this? \n",
    "1922537": "Are you sure? as far as I know they didn't change this interface.. Is there anything else that might be causing your problem?"
  }
}