Berkeley AI Research (BAIR)
UC Berkeley's AI research lab. Pioneering work in robotics, RL, NLP, and computer vision.
bair.berkeley.edu4
483
121
Top Researchers
Recent Papers
The International AI Safety Report 2025's Second Key Update analyzes the current state of AI risk management and technical mitigations employed by researchers, companies, and governments. It highlights advancements in training safer models and monitoring outputs while acknowledging uncertainties in the effectiveness of these measures and their variability across applications. The report aims to inform policymakers, researchers, and the public about progress and remaining gaps in AI safety.
Synthesizes recent developments in AI risk management and technical risk mitigation strategies, identifying both progress and persistent gaps in ensuring the safety of general-purpose AI systems.
The paper introduces EigenSafe, a novel operator-theoretic framework for learning-enabled safety-critical control of stochastic systems by deriving a linear operator governing the dynamic programming principle for safety probability. EigenSafe addresses the limitations of conventional methods like Hamilton-Jacobi reachability and control barrier functions in providing a holistic measure of safety for stochastic robotic systems. They demonstrate that jointly learning the dominant eigenpair of this operator and a safe backup policy offline allows for the construction of a safety filter that detects potentially unsafe situations and reverts to the backup policy, validated across three simulated tasks.
Introduces EigenSafe, a spectral method that learns a safety filter from data by approximating the dominant eigenpair of a linear operator governing safety probability in stochastic systems.
The paper introduces $\pi_{0.5}$, a vision-language-action (VLA) model designed for improved generalization in real-world robotic manipulation tasks. The model builds upon $\pi_{0}$ and employs co-training on heterogeneous data sources, including data from multiple robots, web data, and semantic predictions, to enhance its ability to generalize to unseen environments. Experiments demonstrate that $\pi_{0.5}$ can perform long-horizon, dexterous manipulation skills like cleaning a kitchen or bedroom in novel homes, showcasing the effectiveness of knowledge transfer for real-world robotic systems.
Demonstrates that co-training a VLA model on diverse, heterogeneous data enables effective generalization to long-horizon, dexterous manipulation tasks in unseen real-world environments.
The paper introduces QuantSpec, a self-speculative decoding framework designed to accelerate long-context LLM inference on edge devices by addressing the KV cache bottleneck. QuantSpec employs a draft model that shares the target model's architecture but utilizes a hierarchical 4-bit quantized KV cache and 4-bit quantized weights. The method achieves high acceptance rates (over 90%) and provides consistent end-to-end speedups of up to 2.5x, while also reducing memory requirements by approximately 1.3x compared to other self-speculative decoding methods using sparse KV caches.
Introduces QuantSpec, a self-speculative decoding framework that leverages hierarchical 4-bit quantization of the KV cache and model weights to accelerate long-context LLM inference while maintaining high acceptance rates.