Ask Jeremy anything


(Aseem Bansal) #22

Do you ever get overwhelmed by the amount of new things out there? How do you choose which one to focus your time on?


(Vibhutha Kumarage) #23

How to deploy models that were trained by PyTorch/Fastai into actual production ?.. :slight_smile:


(Nathaniel Shimoni) #24

given the recent work on the fastai library is there any intention to make an end2end autoML pipeline in it?


(Gerardo Garcia) #25

How to deploy models on iOS or Android?
This may be a cool project


(Junxian) #26

Jeremy, I just read this last night, would it be possible for the next fast ai Part3(??), you teach us how to implement Pytorch, Caffe2, in the mobile phone? Thanks!


(Brian Holland) #27

@jeremy How much would you charge for one-on-one help with a project I plan on monetizing?


(Sneha Nagpaul) #28

Could you give us a little hint at how things work with videos or 3d-images and deep learning?


(blake west) #29

Is it doable for students like us (without a PhD) to write deep learning papers that get published? Like do journals review papers from random people? If so, is this kind of thing worth it? (if you’re interested in getting jobs or recognition)


(Aseem Bansal) #30

Would you cover deep learning for audio next time?


(blake west) #31

Any tips or thoughts on getting deep learning jobs in industry?


(Kevin Bird) #32

How does audio modeling compare to Computer Vision or NLP? I was looking for a good audio/video to text dataset (after reading the looking to listen paper) and I couldn’t find anything very useful.


(Brian Holland) #33

How would you recommend getting a job in this field? We can’t all be president of kaggle. :wink:


(blake west) #34

YouTube has transcriptions for tons of videos. Maybe you can get the text and the video from their API?


(Aseem Bansal) #35

https://research.google.com/audioset/


(Kevin Bird) #36

Worth a shot. I will check it out.


(Kevin Bird) #37

Does that actually give the text that is being said? I saw that when researching, but it seemed to just give tags of the video.


(Junxian) #38

And also, it is true that in the real job, 80% is about data cleaning and analysis and only 20% is about machine learning?


(Chan Sooi Loong) #39

How do we extract entity, and key phrases?


(Bharadwaj Srigiriraju) #40

How has competition on Kaggle changed over the years, and what additional skills are needed for a fast.ai student to do well in current Kaggle competitions?


(Aseem Bansal) #41

My bad. I was skimming and just read audio. Not audio to text.