Coherence boosting: When your pretrained language model is not paying enough attention [article]

Nikolay Malkin, Zhen Wang, Nebojsa Jojic
<span title="2022-03-16">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Long-range semantic coherence remains a challenge in automatic language generation and understanding. We demonstrate that large language models have insufficiently learned the effect of distant words on next-token prediction. We present coherence boosting, an inference procedure that increases a LM's focus on a long context. We show the benefits of coherence boosting with pretrained models by distributional analyses of generated ordinary text and dialog responses. It is also found that
more &raquo; ... boosting with state-of-the-art models for various zero-shot NLP tasks yields performance gains with no additional training.
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2110.08294v2">arXiv:2110.08294v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/ggittgqw5farnksaz7ggswlp7i">fatcat:ggittgqw5farnksaz7ggswlp7i</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20211020231926/https://arxiv.org/pdf/2110.08294v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/cc/1d/cc1db851e3881be28564aca2ef0fecae133c45a1.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2110.08294v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>