Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
YjSe2GMQ
on Feb 4, 2019
|
parent
|
context
|
favorite
| on:
Transformer-XL: Unleashing the Potential of Attent...
How well does it do? Would you have a link?
PaulHoule
on Feb 4, 2019
[–]
https://paperswithcode.com/sota/character-level-models-hutte...
terrelln
on Feb 4, 2019
|
parent
[–]
That site claims 0.99 BPC, which I guess means the compressed size is 10^8 * 0.99 / 8 ~= 12,375,000.
Does that include the size of the model, or is this just the cost to encode the errors?
robrenaud
on Feb 4, 2019
|
root
|
parent
[–]
That is just the cost to encode errors.
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: