Lesson 1 In-Class Discussion


@memetzgz, ok. Thanks for your response.


Where are results stored after running the course1 notebook in crestle environment? I like to see the output. (predicted labels for images under ‘test’ directory).

(Sam Lloyd) #273

So I’m running into this:

AssertionError                            Traceback (most recent call last)
<ipython-input-12-d4fe8b8bb391> in <module>()
      3 # Setting requires_grad=False indicates that we do not need to compute gradients
      4 # with respect to these Variables.
----> 5 x = Variable(x.type(dtype), requires_grad=False)
      6 y = Variable(y.type(dtype), requires_grad=False)

/home/ubuntu/anaconda3/lib/python3.5/site-packages/torch/_utils.py in _type(self, new_type, async)
     33     if new_type.is_sparse:
     34         raise RuntimeError("Cannot cast dense tensor to sparse tensor")
---> 35     return new_type(self.size()).copy_(self, async)

/home/ubuntu/anaconda3/lib/python3.5/site-packages/torch/cuda/__init__.py in _lazy_new(cls, *args, **kwargs)
    264 @staticmethod
    265 def _lazy_new(cls, *args, **kwargs):
--> 266     _lazy_init()
    267     # We need this method only for lazy init, so we can remove it
    268     del _CudaBase.__new__

/home/ubuntu/anaconda3/lib/python3.5/site-packages/torch/cuda/__init__.py in _lazy_init()
     82         raise RuntimeError(
     83             "Cannot re-initialize CUDA in forked subprocess. " + msg)
---> 84     _check_driver()
     85     torch._C._cuda_init()
     86     torch._C._cuda_sparse_init()

/home/ubuntu/anaconda3/lib/python3.5/site-packages/torch/cuda/__init__.py in _check_driver()
     56 Found no NVIDIA driver on your system. Please check that you
     57 have an NVIDIA GPU and installed a driver from
---> 58 http://www.nvidia.com/Download/index.aspx""")
     59         else:
     60             # TODO: directly link to the alternative bin that needs install

Found no NVIDIA driver on your system. Please check that you
have an NVIDIA GPU and installed a driver from

I’m running on a p2 instance on AWS, so very much have a GPU! a bit stumped on what to do

(Freddie Vargus) #274

Watching a lecture right now from Yann LeCun :slight_smile:


(Benjamin DeKoven) #275


For how many lessons will Crestle be an acceptable environment?

If not optimum for all lessons, then will another AWS AMI be setup?

I am asking since the current AWS AMI requires a lot of setup adjustment as you are already aware. I do believe that this activity is a great learning step. But I also need to spend effort on lesson1 to start the deep learning mindfulness. The Crestle environment serves this purpose right now for me.

Thank you,

(Ibrahim El-Fayoumi) #276

Python 3.63
It worked for a while and after installing labextension I got this error

(Jeremy Howard) #277

Crestle will be fine for the whole course. I’m setting up an AMI right now, which will not require any extra setup to use.

(Victor Alfonso Arias Vanegas) #278

Thank you Mr. @jeremy for this AMI, i am waiting for this.

(Zao Yang) #279


(Vikrant Behal) #280

@jeremy @yinterian Has the data set for cats vs dogs changed from v1?

(Jeremy Howard) #281


(Chris Palmer) #282

Hi santhaman

As we are running a fastai environment under Anaconda, I am confused about adding missing libraries like OpenCV. Do we add them after activating the fastai environment, or exit to our main Python 3.6 and install them there?

(yinterian) #283

You can install them after you activate the environment or

    conda install -n <env_name>  <package>

If you are in the environment you can also use pip.

(Chris Palmer) #284

Thanks for that advice! If I think it would be a good idea to have those libraries anyway in my core Python system, can I install them back there and they become available whenever I load my fast.ai environment? Or do I have to install twice, in the core system AND the environment?


When loading a particular environment, you can only access the packages installed in that particular environment.

(Daniel Freire) #286

Hey guys! I have a question: what’s the purpose of sz?

(Aditya) #287

IMG Size

(Daniel Freire) #288


(Rahul Pathak) #289

One clarification, here we are using already trained model resnet which was trained on another image dataset and now we are using it to adapt it according to our dog cat dataset: Is this called Transfer Learning or something else

(João Marcos Gris) #290

My version of crestle environment only has the old version of the course.