Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
belval
on Oct 28, 2021
|
parent
|
context
|
favorite
| on:
Our datasets are flawed. ImageNet has an error rat...
Interestingly, these are also used for a lot of CV models that have nothing to do with ImageNet.
cinntaile
on Oct 28, 2021
[–]
I wonder if that's because the numbers are general enough across many different tasks or if normalization just isn't as important as we assume?
Consider applying for YC's Winter 2027 batch!
Applications
are open till November 2.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
