About the Part 1 (2020) category

Part 1 (2020)
107

This just made my morning! Very excited to participate in the new course with v2 coming out :slight_smile:

108

I would never think that I would get invited, it was only a wish :star_struck: Thanks a lot for it! I am happy to be part of this community and looking forward to learning and contributing.

3 Likes
109

Iā€™m psyched about diving back into deep learning! Thanks for this opportunity!

Is there a link to the class schedule? Or an ical event series that we can subscribe to? I want to get this marked off ASAP :slight_smile:

1 Like
110

Thatā€™s very interesting. Iā€™m curious to see your progress with the box.

Iā€™ve trained a Mask R-CNN recently and found it to need considerably more GPU memory (my 7GB only did it for a couple of cycles of Facebookā€™s Detectron2 training - I had to switch to a cloud instance with 16 GB). So in my box Iā€™d want at least 16GB. With your option to include a second GPU at a later point in time you should be fine as well. Have you looked at the NVLink Bridge at all?

Regarding the MKL I have no idea how much that affects the training. I gave the Intelā€™s MKL-DNN in combination with Mask R-CNN (Detectron2 implementation) a spin for inference without GPU support but wasnā€™t impressed by the performance (best inference performance was 3s/image) - I used Google Cloud Platform N2 instances with 4, 8 and 16 vCPU cores. My good old 1070 managed to run one image in 300-500 ms. Might have to do with unsupported / very expensive operations. Anyway, I canā€™t tell how big of a difference either MKL or MKL-DNN make - I would imagine the impact is much lower if you have a GPU onboard.

111

Thatā€™s fantastic! Here we go againā€¦ Thanks @jeremy for the invitation.

112

Thank you again.

113

Oops silly me! Added to the top post. Itā€™s:

Dates: March 17 - May 5 (no class April 21)
Schedule: Tuesdays, 6:30-9:00pm

If someone wants to create an iCal calendar Iā€™ll add it to the top post.

1 Like
114

Yeah a lot of folks have to join in the middle of the night, sadlyā€¦ (You can instead watch the recording the day after, although that means you canā€™t ask questions during the class.)

3 Likes
115

The last course was a direct feed from the in-room audio using a lavelier mic, encoded as 128k mp3. Iā€™m not sure thereā€™s much we can do to improve on that.

116

Yes you will - although you wonā€™t be able to ask questions during class of course if you watch the next day.

1 Like
117

An iCal iCal
The link for Google Calendar

And a post with an embedded calendar:

Hope it helps

7 Likes
118

Thanks Jeremy! That was a very pleasant surprise :slight_smile: Hope to help translating the transcriptions to Portuguese. In the last few months, Iā€™ve been very slowly also translating the FastAIā€™s documentation to my language: https://github.com/WittmannF/fastai-docs-ptbr/blob/master/README.md. For those interested, I created a script for translating jupyter notebook files to any language: https://github.com/WittmannF/jupyter-translate

5 Likes
119

NB: For the calendar Iā€™ve copied the info from the in-person course page (address, content, etc.) Let me know if there are errors in it.

1 Like
120

Thanks for the invitation, it was unexpected and pleasant!
I would not be able to watch live, but hope to help with library :slight_smile:

121

This made my day!
Iā€™m very grateful and looking forward to it, time to get ready during next 7 weeks !!! :slight_smile:

122

My setup consists of a Repulib of Gamers Strix laptop with a mobile GeForce GTX 1070. Itā€™s heavy (and so is the charger), so I usually avoid carrying it around - traveling with two laptops is kind of an overkill :wink: :weight_lifting_man:

123

Thank you Jeremy! Thatā€™s awesome, very excited!

124

Thanks for the invitation, Jeremy!!! Looking forward for the course!

125

Thanks @jeremy , Iā€™m looking forward the course!

Congratulations for writing the book!

126

Say that after weā€™ve actually written the damn thingā€¦ :wink:

9 Likes