Hi there, just wondering, when I train the ResNet model in jupyter notebook. Is that being trained on my laptops CPU, or some virtual machine in the cloud? I was surprised at how fast it ran, especially with so many trainable parameters! I couldn’t see a noticeable change in my PCs utilisation so I’m guessing it’s on a virtual machine?
Thanks in advance.
Unless you have installed Jupyter locally, launched the server locally, and pointed your browser to your local instance when you open the notebook, then it is running on the cloud. What does your URL look like?
Here’s what it looks like if you’re running locally, first the server launch…
conda activate tf_macos
(tf_macos) % jupyter notebook
[I 05:53:13.149 NotebookApp] Serving notebooks from local directory: /Users/xxx/Documents/tf/YOLO
[I 05:53:13.149 NotebookApp] Jupyter Notebook 6.4.4 is running at:
[I 05:53:13.149 NotebookApp] http://localhost:8888/?token=79c4545dc24c0bcf5fa79253ba3f484180e6095db143d82c
[I 05:53:13.149 NotebookApp] or http://127.0.0.1:8888/?token=79c4545dc24c0bcf5fa79253ba3f484180e6095db143d82c
[I 05:53:13.149 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 05:53:13.151 NotebookApp]
and the browser URL…
http://localhost:8888/notebooks/Yolo-TF2-v5.ipynb
Yes, the course notebooks run the Jupyter “kernel”, which is where the actual execution happens, in the cloud. I think all of Coursera runs on AWS basically. The assignments are structured as Docker images, running in linux VMs running on AWS. You’re just running the browser UI locally. You can even use a tablet or phone to run the notebooks, although having a full sized screen and a real keyboard seems like the way to go. 