无法从AWS sagemaker中的操作系统环境获取gpu计数

Question

if __name__ == '__main__':

print(os.environ[ 'JUPYTER_PATH'])


parser = argparse.ArgumentParser() 
args , _ =parser.parse_known_args()

print(os.environ)
parser.add_argument('--training', type=str, default=os.environ['SM_CHANNEL_TRANING'])
parser.add_argument('--gpu-count', type=int, default=os.environ['SM_NUM_GPUS'])


epochs         = 10
lr             = 0.2
batch_size     = 250
gpu_count      = 1
traning_dir    = args.training
#validation_dir = 


train_images = np.load(os.path.join(training_dir, 'traning.npz')[enter image description here][1])['image']
train_labels = np.load(os.path.join(training_dir, 'traning.npz'))['label']
train_images = np.load(os.path.join(validation_dir, 'validation.npz'))['image']
train_images = np.load(os.path.join(validation_dir, 'validation.npz'))['label']


K.set_image_data_format('channels_last')


train_images = train_image.reshape(train_images.shape[0], 32, 32, 3)
test_images = test_image.reshape(train_images.shape[0], 32, 32, 3)
input_shape = (32, 32, 3)


train_images = train_images.astype('float32')
test_images = test_images.astype('float32')
train_images /= 255 
test_images /= 255

train_labels = tenserflow.keras.utils.to.categorical(train_labels, 43)
test_labels =  tenserflow.keras.utils.to.categorical(test_labels, 43)




"""
parser = argparse.ArgumentParser()
parser.add_argument('--epochs', type=int, default=1)
parser.add_argument('--learning-rate', type=float, default=0.001)
parser.add_argument('--batch-sizes', type=int, default=32)

print(os.environ)

parser.add_argument('--gpu-count', type=int, default=os.environ['SM_NUM_GPUS'])
parser.add_argument('--model-dir', type=str, default=os.environ['SM_MODEL_DIR'])
parser.add_argument('--training', type=str, default=os.environ['SM_CHANNEL_TRANING'])
parser.add_argument('--validation', type=str, default=os.environ['SM_CHANNEL_VALIDATION'])


args, _ = parser.parser_known_args()

"""

这是我的错误。

-
KeyError                                  Traceback (most recent call last)
<ipython-input-4-f9bc0f149346> in <module>
      8 
      9     print(os.environ)
---> 10     parser.add_argument('--gpu-count', type=int, default=os.environ['SM_NUM_GPUS'])
     11     parser.add_argument('--training', type=str, default=os.environ['SM_CHANNEL_TRANING'])
     12 

/usr/local/lib/python3.7/os.py in __getitem__(self, key)
    679         except KeyError:
    680             # raise KeyError with the original key value
--> 681             raise KeyError(key) from None
    682         return self.decodevalue(value)
    683 

    **KeyError: 'SM_NUM_GPUS'**

can't find dictionary in the environment

key error，谁能帮帮我？

Answer 1

看起来代码没有从环境中获取值。

您不能运行带有参数--gpu-count的代码，因此代码在操作系统级别签入环境变量。

SM_NUM_GPUS=1

您可以在代码中设置它，并在以下位置检查引用：https://github.com/aws/sagemaker-containers#sm_num_gpus

import os

# using it in argparse
parser.add_argument('num_gpus', type=int, default=os.environ['SM_NUM_GPUS'])

# using it as variable
num_gpus = int(os.environ['SM_NUM_GPUS'])