update pytorch DLC version to 1.11

The notebook fails with current 1.8 pytorch. I think its a problem with the torchvision installed in the container.

```
AlgorithmError: ExecuteUserScriptError: Command "/opt/conda/bin/python3.6 mnist.py --backend gloo --epochs 1" INFO:__main__:Initialized the distributed environment: 'gloo' backend on 2 nodes. Current host rank is 0. Number of gpus: 0 INFO:__main__:Get train data loader Traceback (most recent call last): File "mnist.py", line 257, in <module> train(parser.parse_args()) File "mnist.py", line 114, in train train_loader = _get_train_data_loader(args.batch_size, args.data_dir, is_distributed, **kwargs) File "mnist.py", line 48, in _get_train_data_loader [transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))] File "/opt/conda/lib/python3.6/site-packages/torchvision/datasets/mnist.py", line 83, in __init__ ' You can use download=True to download it') RuntimeError: Dataset not found. You can use download=True to download it, exit code: 1
```

sagemaker-python-sdk/pytorch_mnist/pytorch_mnist.ipynb

@@ -204,8 +204,8 @@
     "\n",
     "estimator = PyTorch(entry_point='mnist.py',\n",
     "                    role=role,\n",
-    "                    py_version='py3',\n",
-    "                    framework_version='1.8.0',\n",
+    "                    py_version='py38',\n",
+    "                    framework_version='1.11.0',\n",
     "                    instance_count=2,\n",
     "                    instance_type='ml.c5.2xlarge',\n",
     "                    hyperparameters={\n",