Timezone: »

PLATON: Pruning Large Transformer Models with Upper Confidence Bound of Weight Importance
Qingru Zhang · Simiao Zuo · Chen Liang · Alexander Bukharin · Pengcheng He · Weizhu Chen · Tuo Zhao

Wed Jul 20 03:30 PM -- 05:30 PM (PDT) @ Hall E #230

Large Transformer-based models have exhibited superior performance in various natural language processing and computer vision tasks. However, these models contain enormous amounts of parameters, which restrict their deployment to real-world applications. To reduce the model size, researchers prune these models based on the weights' importance scores. However, such scores are usually estimated on mini-batches during training, which incurs large variability/uncertainty due to mini-batch sampling and complicated training dynamics. As a result, some crucial weights could be pruned by commonly used pruning methods because of such uncertainty, which makes training unstable and hurts generalization. To resolve this issue, we propose PLATON, which captures the uncertainty of importance scores by upper confidence bound of importance estimation. In particular, for the weights with low importance scores but high uncertainty, PLATON tends to retain them and explores their capacity. We conduct extensive experiments with several Transformer-based models on natural language understanding, question answering and image classification to validate the effectiveness of PLATON. Results demonstrate that PLATON manifests notable improvement under different sparsity levels. Our code is publicly available at https://github.com/QingruZhang/PLATON.

Author Information

Qingru Zhang (Georgia Institute of Technology)

Qingru Zhang is a Ph.D. student at Georgia Tech. His research mainly focuses on developing principled learning algorithms with an emphasis on language models and graph representation learning.

Simiao Zuo (Georgia Institute of Technology)
Chen Liang (Georgia Institute of Technology)
Alexander Bukharin (Georgia Institute of Technology)
Pengcheng He (Microsoft)
Weizhu Chen (Microsoft)
Tuo Zhao (Georgia Tech)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors