We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Discourse-Aware Soft Prompting for Text Generation

Abstract: Current efficient fine-tuning methods (e.g., adapters, prefix-tuning, etc.) have optimized conditional text generation via training a small set of extra parameters of the neural language model, while freezing the rest for efficiency. While showing strong performance on some generation tasks, they don't generalize across all generation tasks. We show that soft-prompt based conditional text generation can be improved with simple and efficient methods that simulate modeling the discourse structure of human written text. We investigate two design choices: First, we apply \textit{hierarchical blocking} on the prefix parameters to simulate a higher-level discourse structure of human written text. Second, we apply \textit{attention sparsity} on the prefix parameters at different layers of the network and learn sparse transformations on the softmax-function. We show that structured design of prefix parameters yields more coherent, faithful and relevant generations than the baseline prefix-tuning on all generation tasks.
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2112.05717 [cs.CL]
  (or arXiv:2112.05717v2 [cs.CL] for this version)

Submission history

From: Marjan Ghazvininejad [view email]
[v1] Fri, 10 Dec 2021 18:15:44 GMT (3819kb,D)
[v2] Mon, 23 May 2022 17:27:22 GMT (6410kb,D)

Link back to: arXiv, form interface, contact.