In-context learning (ICL, also known as few-shot prompting) has been the standard method of adapting LLMs to downstream tasks, by learning from a few input-output examples. Nonetheless, all ICL-based approaches only learn from correct input-output pairs. In this paper, we revisit this paradigm, by learning more from the few given input-output examples. We introduce Learning Principles (LEAP): First, we intentionally induce the model to make mistakes on these few examples; then we reflect on these mistakes, and learn explicit task-specific “principles” from them, which help solve similar problems and avoid common mistakes; finally, we prompt the model to answer unseen test questions using the original few-shot examples and these learned general principles. We evaluate LEAP on a wide range of benchmarks, including multi-hop question answering (Hotpot QA), textual QA (DROP), Big-Bench Hard reasoning, and math problems (GSM8K and MATH); in all these benchmarks, LEAP improves the strongest available LLMs such as GPT-3.5-turbo, GPT-4, GPT-4-turbo and Claude-2.1. For example, LEAP improves over the standard few-shot prompting using GPT-4 by 7.5% in DROP, and by 3.3% in HotpotQA. Importantly, LEAP does not require any more input or examples than the standard few-shot prompting settings.

In support of Cloud Robotics, Robotics and Automation as a Service (RAaaS) frameworks have the potential to reduce the complexity of software development, simplify software installation and maintenance, and facilitate data sharing for machine learning. In this proof-of-concept paper, we describe Berkeley Robotics and Automation as a Service (Brass), a RAaaS prototype that allows robots to access a remote server that hosts a robust grasp-planning system (Dex-Net 1.0) that maintains data on hundreds of candidate grasps on thousands of 3D object meshes and uses perturbation sampling to estimate and update a stochastic robustness metric for each grasp. Results suggest that such a system can increase grasp reliability over naive locally-computed grasping strategies with network latencies of 30 and 200 msec for servers 500 and 6000 miles away, respectively. We also study how the system can use execution reports from robots in the field to update grasp recommendations over time.