Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
Der_Einzige
30 days ago
|
parent
|
context
|
favorite
| on:
Qwen 3.8 27B
Min-p is specifically "Don't select tokens less probable than a multiple of the top token's probability" with min_p of 0.1 multiplied by an example top probability of 0.3 being 0.03 as the truncation at that time step.
Source: One of the min_p authors
suprjami
30 days ago
[–]
Thanks, today I learned!
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
Source: One of the min_p authors