Error-driven Pruning of Language Models for Virtual Assistants
This work addresses memory and latency issues in virtual assistants, offering incremental improvements in model efficiency and accuracy.
The paper tackled the problem of large language models for virtual assistants causing memory and real-time issues by customizing entropy pruning with a keep list for infrequent n-grams, resulting in an 8% average Word Error Rate reduction on a targeted test set but with a model three times larger than the baseline.
Language models (LMs) for virtual assistants (VAs) are typically trained on large amounts of data, resulting in prohibitively large models which require excessive memory and/or cannot be used to serve user requests in real-time. Entropy pruning results in smaller models but with significant degradation of effectiveness in the tail of the user request distribution. We customize entropy pruning by allowing for a keep list of infrequent n-grams that require a more relaxed pruning threshold, and propose three methods to construct the keep list. Each method has its own advantages and disadvantages with respect to LM size, ASR accuracy and cost of constructing the keep list. Our best LM gives 8% average Word Error Rate (WER) reduction on a targeted test set, but is 3 times larger than the baseline. We also propose discriminative methods to reduce the size of the LM while retaining the majority of the WER gains achieved by the largest LM.