I'll go first: I remember being really enamored by a model's capacity, so much so that I would frequently burn excess compute time just enjoying the experience of watching my training loss to decline dramatically *even though I understood that wasn't what I cared about.*
-
-
Show this thread
-
Oh, also, I used to spend a lot of time looking for problems that I thought people couldn’t do, only machines could. That’s a really foolish filter. Now I look for problems that even a poorly trained person could do but would be bored as fuck doing yet still must be done.
Show this thread
End of conversation
New conversation -
-
-
There was a time when I had no idea what "Good Enough" meant. Decisions gonna get made, with or without me.
-
Oh yea, I was so bad at this for a while. It *really* is challenging, knowing what is and is not vital. But your's really links up with Vicki's because what you think and what the person who uses it thinks diverge...often quickly.https://twitter.com/vboykis/status/1135599581310464001 …
- 2 more replies
New conversation -
-
-
I have spent more time deploying a model instead of building. Probably around 60-70%
-
- 2 more replies
New conversation -
-
-
doing data science without a strong fundamental base in product and marketing will limit your effectiveness to particular company sizes or particular technical roles
- 3 more replies
New conversation -
-
-
Thinking that messy data was only something that happened in academia and that in industry data lived in tidy databases



-
By replying to this, I discovered new
@rstudio people on here. It's not helping with my prior envy!https://twitter.com/generativist/status/1125797632633331712 …
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.