Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
aydyn
5 months ago
|
parent
|
context
|
favorite
| on:
Mistral AI Launches New 8x22B MOE Model
AFAIK, 2-bit quant leads to too much loss of performance, such that you're better off using a different smaller model altogether. See here:
https://www.reddit.com/r/LocalLLaMA/comments/18ituzh/mixtral...
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
https://www.reddit.com/r/LocalLLaMA/comments/18ituzh/mixtral...