2 papers
cs.IR2024
RE-AdaptIR: Improving Information Retrieval through Reverse Engineered Adaptation
William Fleshman, Benjamin Van Durme
Large language models (LLMs) fine-tuned for text-retrieval have demonstrated state-of-the-art results across several information retrieval (IR) benchmarks. However, supervised trai…
cs.CL2023
Toucan: Token-Aware Character Level Language Modeling
William Fleshman, Benjamin Van Durme
Character-level language models obviate the need for separately trained tokenizers, but efficiency suffers from longer sequence lengths. Learning to combine character representatio…