Bo Liu

Forget complex continual learning strategies: simply fine-tuning large vision-language-action models with LoRA is surprisingly effective for lifelong RL.

Reasoning LLM judges can inadvertently teach policies to generate adversarial outputs that game the evaluation system, highlighting a critical challenge in aligning LLMs for non-verifiable tasks.

Forget everything you thought you knew about continual learning: pretrained Vision-Language-Action models can learn new robotic skills without catastrophic forgetting, even with minimal replay.

You can now get accurate time-series forecasts with feature-level interpretability, enabling simpler and more efficient early warning systems.

Massively parallelizing multi-task RL reveals unexpected challenges, suggesting that simply scaling up existing algorithms may not be sufficient for optimal performance in complex robotics scenarios.