NeurIPS edition

:partying_face: We’re thrilled to be an official participant of NeurIPS meetups, where your host Sanyam will be summarizing the top 5 papers presented at NeurIPS 2021 and presenting their impact, background and gist to the audience.

:books: Join us live here: wandb.me/prg

3 Likes

Q: How is this better than GBT?

Tabular data is what Industry uses most. Even if you think you are doing Image, video, and texts, there’s always Metadata associated, which is Tabular data (in CSV or JSON or other formats).

@bhutanisanyam1 - Can you post the links to papers we are discussing here?

Hi Sanyam, sorry if I have missed this, did the author also said that Resnet and trasnformer (both or any one) gave better performances than boosted algorithms?

More people on YouTube LiveSteam than on Zoom? Is it because Meetup Link for this points to LiveStream instead of Zoom? :thinking:

@bhutanisanyam1 In the earlier paper on MLPs, does the author compare gMLPs with larger Transformers as well? Like BERT Large for instance

1 Like

Thanks for every thing that you do Sanyam to learn and to inspire interest. Francois Cholet one will be amazing!! Recently brought the book!! Yay.

1 Like

And did I heard you correctly that JAX group will continue? I thought you said yesterday that this the last session.

1 Like

If you are planning to go in deeper in one of the paper in wandb’s next paper read session, can I request you to consider the K-Net one? Also, whenever you get time please share the paper and repo links. Thanks.

1 Like