Timezone: »

 
A Fast, Fisher Based Pruning of Transformers without Retraining
Amir Gholami

Fri Dec 02 02:00 PM -- 02:45 PM (PST) @
in Order up! The Benefits of Higher-Order Optimization in Machine Learning »

Pruning is an effective way to reduce the huge inference cost of large Transformer models. However, prior work on model pruning requires retraining the model. This can add high cost and complexity to model deployment, making it difficult to use in many practical situations. To address this, we propose a fast post- training pruning framework for Transformers that does not require any retraining. Given a resource constraint and a sample dataset, our framework automatically prunes the Transformer model using structured sparsity methods. To retain high accuracy without retraining, we introduce three novel techniques: (i) a lightweight mask search algorithm that finds which heads and filters to prune based on the Fisher information; (ii) mask rearrangement that complements the search algorithm; and (iii) mask tuning that reconstructs the output activations for each layer. We apply our method to BERT-BASE and DistilBERT, and we evaluate its effectiveness on GLUE and SQuAD benchmarks. Our framework achieves up to 2.0x reduction in FLOPs and 1.56x speedup in inference latency, while maintaining < 1\% loss in accuracy. Importantly, our framework prunes Transformers in less than 3 minutes on a single GPU, which is over two orders of magnitude faster than existing pruning approaches that retrain.

Author Information

Amir Gholami (University of California, Berkeley)

More from the Same Authors