3 ms·
This doesn't seem to really address synthetic data, let alone RL-based reasoning.
by Philpax 2y ago
This doesn't seem to really address synthetic data, let alone RL-based reasoning.
- cratermoon 2y agoWhy would it? Once those are introduced, advancement leaves behind pure scaling.