Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
ASalazarMX
3 hours ago
|
parent
|
context
|
favorite
| on:
The Tragedy of Google Books (2017)
You punish it if parts of the answer can be found in its training data, and reward it otherwise.
amelius
1 hour ago
[–]
But the whole point of the training is that you reward it if it correctly reproduces the next token.
reply
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: