NAIL: Lexical Retrieval Indices with Efficient Non-Autoregressive Decoders
This work addresses the cost and feasibility issues of deploying neural rerankers for information retrieval, making them more accessible without dedicated hardware.
The paper tackled the problem of high serving costs for neural document rerankers by introducing a lexicalized scoring function that captures up to 86% of the gains of a Transformer model while using only 10-6% of the FLOPs per document and enabling serving on commodity CPUs, and it matches the quality of a state-of-the-art dual encoder retriever when combined with BM25.
Neural document rerankers are extremely effective in terms of accuracy. However, the best models require dedicated hardware for serving, which is costly and often not feasible. To avoid this serving-time requirement, we present a method of capturing up to 86% of the gains of a Transformer cross-attention model with a lexicalized scoring function that only requires 10-6% of the Transformer's FLOPs per document and can be served using commodity CPUs. When combined with a BM25 retriever, this approach matches the quality of a state-of-the art dual encoder retriever, that still requires an accelerator for query encoding. We introduce NAIL (Non-Autoregressive Indexing with Language models) as a model architecture that is compatible with recent encoder-decoder and decoder-only large language models, such as T5, GPT-3 and PaLM. This model architecture can leverage existing pre-trained checkpoints and can be fine-tuned for efficiently constructing document representations that do not require neural processing of queries.