Hacker News new | ask | show | jobs
by chalst 806 days ago
The OP explicitly excludes training.
1 comments

The one I replied to said 1-bit for both training and inference.