Self-Play Pretraining with Zero Data
Advances in language modeling have been driven by scaling pretraining on ever more data. Yet, the training data is still largely curated on the model's behalf. A more general approach to pretraining would let the model learn to generate the data most useful for its own improvement. This would provide an effectively unb...
Aditya Cowsik, Kfir Dolev, Michael Y. Li et al.
· 1 citation