Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
astrange
3 months ago
|
parent
|
context
|
favorite
| on:
Training a trillion parameter model to be funny
The GPT3 base model was pretty funny if you like nonsense. Instruct tuning and RLHF seem to destroy it when they recalibrate everything.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
