Evaluating the Effectiveness of Efficient Neural Architecture Search for Sentence-Pair Tasks
This work addresses the effectiveness of NAS for NLP practitioners, but it is incremental as it tests an existing method on new tasks with limited gains.
The study applied Efficient Neural Architecture Search (ENAS) to sentence-pair tasks like paraphrase detection and semantic textual similarity, finding that ENAS architectures sometimes outperformed LSTMs but performed similarly to random search, with mixed results across datasets and models.
Neural Architecture Search (NAS) methods, which automatically learn entire neural model or individual neural cell architectures, have recently achieved competitive or state-of-the-art (SOTA) performance on variety of natural language processing and computer vision tasks, including language modeling, natural language inference, and image classification. In this work, we explore the applicability of a SOTA NAS algorithm, Efficient Neural Architecture Search (ENAS) (Pham et al., 2018) to two sentence pair tasks, paraphrase detection and semantic textual similarity. We use ENAS to perform a micro-level search and learn a task-optimized RNN cell architecture as a drop-in replacement for an LSTM. We explore the effectiveness of ENAS through experiments on three datasets (MRPC, SICK, STS-B), with two different models (ESIM, BiLSTM-Max), and two sets of embeddings (Glove, BERT). In contrast to prior work applying ENAS to NLP tasks, our results are mixed -- we find that ENAS architectures sometimes, but not always, outperform LSTMs and perform similarly to random architecture search.