Fetching the paper…

BRAIn: Bayesian Reward-conditioned Amortized Inference for natural language generation from feedback · Around