Y
Hacker News
new
|
ask
|
show
|
jobs
by
kadushka
251 days ago
Mainly because global video data corpus is > 100k larger than global text corpus, so you will need to train much larger models for much longer (than current LLMs).