forecast
Text pretraining late innings, multimodal early
Scaling on text pretraining is in the late innings because the internet's text has largely been used, while pretraining on video, image and audio is in the early innings because multimodal data is far more expensive to collect and process.
forecastpending
Speaker
Dylan PatelEvidence Quote
“There's tons of video and image and audio—it's just so expensive. So we didn't get to that.”
Created: 11/16/2026, 4:01:45 AM
Your notes
Loading notes...