Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts

About

The remarkable success of pretrained language models has motivated the study of what kinds of knowledge these models learn during pretraining. Reformulating tasks as fill-in-the-blanks problems (e.g., cloze tests) is a natural approach for gauging such knowledge, however, its usage is limited by the manual effort and guesswork required to write suitable prompts. To address this, we develop AutoPrompt, an automated method to create prompts for a diverse set of tasks, based on a gradient-guided search. Using AutoPrompt, we show that masked language models (MLMs) have an inherent capability to perform sentiment analysis and natural language inference without additional parameters or finetuning, sometimes achieving performance on par with recent state-of-the-art supervised models. We also show that our prompts elicit more accurate factual knowledge from MLMs than the manually created prompts on the LAMA benchmark, and that MLMs can be used as relation extractors more effectively than supervised relation extraction models. These results demonstrate that automatically generated prompts are a viable parameter-free alternative to existing probing methods, and as pretrained LMs become more sophisticated and capable, potentially a replacement for finetuning.

Taylor Shin, Yasaman Razeghi, Robert L. Logan IV, Eric Wallace, Sameer Singh• 2020

Related benchmarks

TaskDatasetResultRank
Topic ClassificationAG-News
Accuracy65.7
225
Text ClassificationSST-2 (test)
Accuracy75
185
Sentiment AnalysisSST-2
Accuracy75
165
Text ClassificationMR (test)
Accuracy62
148
Subjectivity ClassificationSubj (test)
Accuracy78.9
127
Token-forcing loss optimizationRandom targets Held-out (val)
Qwen-2.5-7B Loss11.67
56
JailbreakingHARMBENCH 159 standard behaviors (test)
ASR14.5
51
JailbreakHarmBench Standard Behaviours (200 examples)
ASR0.00e+0
48
Topic ClassificationYahoo (test)
Accuracy35.5
36
Text ClassificationYelp P. (test)
Accuracy79.8
34
Showing 10 of 21 rows

Other info

Follow for update