Prompting has shown impressive success in enabling large pretrained language models (LMs) to perform diverse NLP tasks, especially when only few downstream data are available. Automatically finding the optimal prompt for each task, however, is challenging. Most existing work resorts to tuning soft prompt (e.g., embeddings) which falls short of interpretability, reusability across LMs, and applicability when gradients are not accessible. Discrete prompt, on the other hand, is difficult to optimize, and is often created by "enumeration (e.g., paraphrasing)-then-selection" heuristics that do not explore the prompt space systematically. This paper proposes RLPROMPT, an efficient discrete prompt optimization approach with reinforcement learning (RL). RL-PROMPT formulates a parameter-efficient policy network that generates the desired discrete prompt after training with reward. To overcome the complexity and stochasticity of reward signals by the large LM environment, we incorporate effective reward stabilization that substantially enhances the training efficiency. RLPROMPT is flexibly applicable to different types of LMs, such as masked (e.g., BERT) and left-to-right models (e.g., GPTs), for both classification and generation tasks. Experiments on few-shot classification and unsupervised text style transfer show superior performance over a wide range of existing finetuning or prompting methods. Interestingly, the resulting optimized prompts are often ungrammatical gibberish text; and surprisingly, those gibberish prompts are transferrable between different LMs to retain significant performance, indicating LM prompting may not follow human language patterns. 1
In order to guarantee the liability and accuracy of ultrasonic flaw detectors, a system is developed for measuring the characteristics of ultrasonic flaw detectors. The parameters which evaluate the characteristics of the pulse excitation circuit include pulse amplitude voltage, pulse rise time, pulse duration, pulse repetition frequency, peak frequency, bandwidth and output impedance. The parameters which evaluate the characteristics of the amplifier circuit include cut-off frequency, pass band, dynamic range, input impedance and equivalent input noise. The parameters which evaluate the coupling characteristics of these two circuits include crosstalk and blocking time during transmission.
Acoustic field characteristic is a key factor of the ultrasonic transducer and many acoustic field measuring methods have been studied. However, nowadays hydrophone method which used to measure the acoustical field characteristic of ultrasonic transducer has been wildly used and accepted. This paper introduces an advanced measuring method to get the acoustic field characteristic of the phased array transducer by means of using hydrophone. A new acoustical field scanning method and a synchronization guarantee technique will be introduced. Some experiments have been performed and some interesting phenomena have been found. Then these phenomena have been analyzed and some conclusions have been infered.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.