A year ago I made the resolution to exercise and get more seriously into deep learning. I think I managed to keep one of them and I wrote a little bit about it https://t.co/2j9cKwjpMR
— Sylvain Gugger (@GuggerSylvain) January 2, 2019
A year ago I made the resolution to exercise and get more seriously into deep learning. I think I managed to keep one of them and I wrote a little bit about it https://t.co/2j9cKwjpMR
— Sylvain Gugger (@GuggerSylvain) January 2, 2019
Wrapping up 2018. https://t.co/6d00M95bhE
— Nathan Yau (@flowingdata) December 31, 2018
This is a *must read* for data journalists and data sci folks in general (both groups intersect and have a "shiny object" problem) // When a tech company pulls the plug on your story https://t.co/I2XAFdF0Us
— boB Rudis (@hrbrmstr) December 31, 2018
The more I build, the more I have empathy for builders. https://t.co/QyZEyRZ3Sm
— Chris Albon (@chrisalbon) December 31, 2018
Even though these stories seem small, it is because they repeat themselves and accumulate that I feel so worn down. I am genuinely tired. I think about leaving the field [of machine learning] every other month. https://t.co/ewLWC5pZS9
— Rachel Thomas (@math_rachel) December 30, 2018
As long as the paper says "work in progress", the reader is warned.
— Yann LeCun (@ylecun) December 29, 2018
Anything that will accelerate the exchange of ideas is good, IMO.
Economists have "working papers".
My understanding is that it’s done to lock in credit for completing important steps that are believed to be part of an eventual proof
— Ian Goodfellow (@goodfellow_ian) December 29, 2018
I see a lot of @arxiv papers marked as "work in progress" or "ongoing paper draft". The purpose of arXiv is to publish preprints for papers that have been submitted and/or accepted for publication. It is not a place to checkpoint your drafts.
— Thomas G. Dietterich (@tdietterich) December 29, 2018
This https://t.co/x6hxbnkC6h pic.twitter.com/bOrqAfKMdP
— Delip Rao (@deliprao) December 29, 2018
If you don't know what type a function expects you don't know how to use it - so it's critical documentation for users. We've also set up our doc building framework so that all fastai and pytorch types are auto-linked to their documentation, eg: https://t.co/kXio6x9OnF
— Jeremy Howard (@jeremyphoward) December 29, 2018
Rasa open source chatbot API: Enterprises & customers want AI assistants, not FAQ chatbots, but they’re difficult to build; Since Google demoed Duplex, every developer, product manager, and executive wants their own that can handle contextual conversations https://t.co/MxNxaHAasd
— Stanford NLP Group (@stanfordnlp) December 29, 2018
reminder: not all mentorship is the same https://t.co/NkVc5Yseo9 pic.twitter.com/rsel6Emg6W
— Rachel Thomas (@math_rachel) December 28, 2018