Single-Pass Document Scanning for Question Answering
This provides a more efficient solution for question answering over massive text, addressing a bottleneck in handling long documents for researchers and practitioners in NLP.
The paper tackles the challenge of processing extremely large documents for question answering by proposing a single-pass document scanning approach that preserves global coherence and processes text in linear time. It outperforms chunk-based embedding methods on 41 QA benchmarks and competes with large language models at a fraction of the computational cost.
Handling extremely large documents for question answering is challenging: chunk-based embedding methods often lose track of important global context, while full-context transformers can be prohibitively expensive for hundreds of thousands of tokens. We propose a single-pass document scanning approach that processes the entire text in linear time, preserving global coherence while deciding which sentences are most relevant to the query. On 41 QA benchmarks, our single-pass scanner consistently outperforms chunk-based embedding methods and competes with large language models at a fraction of the computational cost. By conditioning on the entire preceding context without chunk breaks, the method preserves global coherence, which is especially important for long documents. Overall, single-pass document scanning offers a simple solution for question answering over massive text. All code, datasets, and model checkpoints are available at https://github.com/MambaRetriever/MambaRetriever