General course chat

(Paulina Herra) #45

Thanks a bunch, @leac! I knew I read it somewhere, I just couldn’t find it :slight_smile:

0 Likes

(Ronildo Moura) #47

thank you sir.

0 Likes

(Pawan Sasanka) #48

@rachel @jeremy if i’m writing a blog post on this, Can i use the notebook link i.e the fast ai v3 repo or should i just individually use the code and respective docs?

1 Like

(Jeremy Howard (Admin)) #50

Perhaps post a draft here first of your post?

1 Like

(Jeremy Howard (Admin)) split this topic #51

8 posts were split to a new topic: Deconvolution issue

0 Likes

(Ryan Kirkman) #52

I don’t currently have permission to edit this wiki post (perhaps this account needs more trust level?)

The Platform: Sagemaker link currently points to: https://forums.fast.ai/t/lesson-1-class-discussion-and-resources/27332/3
Could we please point it at: https://forums.fast.ai/t/platform-sagemaker/27333 ?

0 Likes

(Jeremy Howard (Admin)) #53

My fault for forgetting to make it a wiki. Fixed now. Do you mind trying again?

1 Like

(Ryan Kirkman) #54

Works perfectly now, thanks Jeremy.

0 Likes

(Adam Reevesman) #55

I am looking into the installation instructions for Google Compute Platform. Specifically, I followed the link in “To install on MacOS, follow the instructions detailed in points 1 to 4 here” under “Step 2: Install Google CLI” on Mac.

The first step under “Before you Begin” says to make sure python 2.7 is installed on your system, but there is also a note saying " Note: As of Cloud SDK version 206.0.0, the gcloud CLI has experimental support for running using a Python 3.4+ interpreter (run gcloud topic startup for exclusions and more information on configuring your Python interpreter). All other Cloud SDK tools still require a Python 2.7 interpreter."

I am using python 3.6. In light of the note will I be okay to use Google Compute Platform anyway?

Apologies for already posting this question once in “Lesson 1: Class Discussion and Resources,” but I learned that this is the more appropriate thread.

0 Likes

(Francisco Ingham) #56

I am sorry about the confusion. Could you try this guide instead with Python 3.6?

I think it should work but let me know if it doesn’t.

0 Likes

(hector) #57

@areevesman I think this is the correct thread https://forums.fast.ai/t/platform-gcp/27375?u=raghavab1992 to get this adressed

1 Like

#59

Wasn’t sure where to mention this but http://course-v3.fast.ai/start_gcp.html

Step 4: the file dir seems to have changed and this:

cd tutorials/fastai/course-v3
git checkout .
git pull

…no longer works and as such is obv no longer a git repo.

Would be good to get a steer also on where I should mention these corrections going forward if this is wrong place :smile:

0 Likes

(Jeremy Howard (Admin)) #60

Yup fixed that earlier today - thanks. Better places to mention are https://forums.fast.ai/t/platform-gcp/27375 , or best is a PR to https://github.com/fastai/course-v3/

1 Like

#61

I am trying to run the pets-notebook and I encounter the following error:

----> 1 learn = ConvLearner(data, models.resnet34, metrics=error_rate) AttributeError: module 'fastai.vision.models' has no attribute 'resnet34'


Any ideas?

1 Like

#62

This error is speficially mentioned in the FAQ. Don’t forget to check it, or do a quick search on the forum before posting!

1 Like

#63

Any chance you might include a lesson on Text to Speech? I find Deep Voice 3 fascinating

I’ve been try out a couple of open source implementations which work pretty, it would be nice to understand more about how they work…


5 Likes

(Jeremy Howard (Admin)) #64

Yes I’d like to in part 2.

13 Likes

(Theodoros Galanos) #65

FAIR released faster R-CNN and Mask R-CNN for pytorch 1.0, if anyone is interested. Good place to take a look until we go over it in the course (I’d guess part 2).

4 Likes

(George Zhang) #66

Try to look and see if there is a course-v3 folder under under the home directory with ls.

0 Likes

#67

understood - thanks for clarification.

0 Likes