4 ms·
The market for using multi head attention based transformers is firming up fast, but it's one algorithmic innovation away from all those foundation models being
by nightski 3y ago
The market for using multi head attention based transformers is firming up fast, but it's one algorithmic innovation away from all those foundation models being worthless and starting over from scratch.