# ReBRAC-v2: The Return of the King

arXiv: https://arxiv.org/abs/2608.01205 · HTML: https://mlsanity.com/paper/2608.01205/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Offline RL algorithms, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 27 of its 32 references that are in the map.
- First posted on arXiv 2 Aug 2026.

## Placed in

- Offline RL and imitation learning from data / [Offline RL algorithms](https://mlsanity.com/offlinerl/offline-rl-algorithms/)

## Builds on (most cited first, 27 of 27)

- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Addressing Function Approximation Error in Actor-Critic Methods](https://mlsanity.com/paper/1802.09477/) (2018)
- [Variational Inference with Normalizing Flows](https://mlsanity.com/paper/1505.05770/) (2015)
- [Density estimation using Real NVP](https://mlsanity.com/paper/1605.08803/) (2016)
- [Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems](https://mlsanity.com/paper/2005.01643/) (2020)
- [Glow: Generative Flow with Invertible 1x1 Convolutions](https://mlsanity.com/paper/1807.03039/) (2018)
- [D4RL: Datasets for Deep Data-Driven Reinforcement Learning](https://mlsanity.com/paper/2004.07219/) (2020)
- [Offline Reinforcement Learning with Implicit Q-Learning](https://mlsanity.com/paper/2110.06169/) (2021)
- [A Minimalist Approach to Offline Reinforcement Learning](https://mlsanity.com/paper/2106.06860/) (2021)
- [Behavior Regularized Offline Reinforcement Learning](https://mlsanity.com/paper/1911.11361/) (2019)
- [Neural Spline Flows](https://mlsanity.com/paper/1906.04032/) (2019)
- [Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble](https://mlsanity.com/paper/2110.01548/) (2021)
- [OGBench: Benchmarking Offline Goal-Conditioned RL](https://mlsanity.com/paper/2410.20092/) (2024)
- [Adan: Adaptive Nesterov Momentum Algorithm for Faster Optimizing Deep Models](https://mlsanity.com/paper/2208.06677/) (2022)
- [Stop Regressing: Training Value Functions via Classification for Scalable Deep RL](https://mlsanity.com/paper/2403.03950/) (2024)
- [CORL: Research-oriented Deep Offline Reinforcement Learning Library](https://mlsanity.com/paper/2210.07105/) (2022)
- [Dropout Q-Functions for Doubly Efficient Reinforcement Learning](https://mlsanity.com/paper/2110.02034/) (2021)
- [Is Value Learning Really the Main Bottleneck in Offline RL?](https://mlsanity.com/paper/2406.09329/) (2024)
- [Offline Q-Learning on Diverse Multi-Task Data Both Scales And Generalizes](https://mlsanity.com/paper/2211.15144/) (2022)
- [Improving Regression Performance with Distributional Losses](https://mlsanity.com/paper/1806.04613/) (2018)
- [Let Offline RL Flow: Training Conservative Agents in the Latent Space of Normalizing Flows](https://mlsanity.com/paper/2211.11096/) (2022)
- [Butterfly Effects of SGD Noise: Error Amplification in Behavior Cloning and Autoregression](https://mlsanity.com/paper/2310.11428/) (2023)
- [Learning to Optimize for Reinforcement Learning](https://mlsanity.com/paper/2302.01470/) (2023)
- [NinA: Normalizing Flows in Action. Training VLA Models with Normalizing Flows](https://mlsanity.com/paper/2508.16845/) (2025)
- [Mining Generalizable Activation Functions](https://mlsanity.com/paper/2602.05688/) (2026)
- [SERNF: Sample-Efficient Real-World Dexterous Policy Fine-Tuning via Action-Chunked Critics and Normalizing Flows](https://mlsanity.com/paper/2602.09580/) (2026)
- [The Role of Deep Learning Regularizations on Actors in Offline RL](https://mlsanity.com/paper/2409.07606/) (2024)

## Abstract (arXiv)

Recent offline reinforcement learning methods increasingly rely on expressive generative policies and specialized value-guidance mechanisms. We ask whether comparable progress can instead come from systematically modernizing a conventional behavior-regularized actor-critic while preserving its algorithmic simplicity. We introduce ReBRAC-v2, which directly trains an exact-likelihood normalizing flow as the RL actor, combines likelihood, MSE, and MAE behavior regularization, and integrates a classification-based residual critic, staged optimization, and multi-sample test-time action selection. Rather than tuning this recipe separately for every task, we develop a single shared configuration via roughly 600 Bayesian proposals on six challenging OGBench tasks, freeze all structural and optimization choices, and adapt only two behavior-regularization coefficients over a 16-point grid. Across ten common state-based OGBench categories, ReBRAC-v2 averages 74.8 compared to 52.3 for the next-best aggregate result and ranks first in eight categories. The same recipe, without structural changes, obtains the strongest averages in our comparisons on D4RL AntMaze (90.2) and Adroit (33.6). Fixed-recipe ablations show the largest sensitivity to the selected mixed cloning objective, staged training, sufficient flow capacity, and multi-sample inference, while showing that several smaller choices depend on the values of other hyperparameters. These results show that disciplined, transferable engineering can achieve state-of-the-art aggregate performance without abandoning a minimalist offline RL foundation.