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

Download:

Current browse context:

cs.CL

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(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: Continuous diffusion for categorical data

Abstract: Diffusion models have quickly become the go-to paradigm for generative modelling of perceptual signals (such as images and sound) through iterative refinement. Their success hinges on the fact that the underlying physical phenomena are continuous. For inherently discrete and categorical data such as language, various diffusion-inspired alternatives have been proposed. However, the continuous nature of diffusion models conveys many benefits, and in this work we endeavour to preserve it. We propose CDCD, a framework for modelling categorical data with diffusion models that are continuous both in time and input space. We demonstrate its efficacy on several language modelling tasks.
Comments: 26 pages, 8 figures; corrections and additional information about hyperparameters
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2211.15089 [cs.CL]
  (or arXiv:2211.15089v3 [cs.CL] for this version)

Submission history

From: Sander Dieleman [view email]
[v1] Mon, 28 Nov 2022 06:08:54 GMT (350kb,D)
[v2] Tue, 6 Dec 2022 15:52:02 GMT (350kb,D)
[v3] Thu, 15 Dec 2022 14:27:19 GMT (350kb,D)

Link back to: arXiv, form interface, contact.