Towards Latent Diffusion Suitable For Text
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908783176843264 |
|---|---|
| author | Midavaine, Nesta Naesseth, Christian A. Bartosh, Grigory |
| author_facet | Midavaine, Nesta Naesseth, Christian A. Bartosh, Grigory |
| contents | Language diffusion models aim to improve sampling speed and coherence over autoregressive LLMs. We introduce Neural Flow Diffusion Models for language generation, an extension of NFDM that enables the straightforward application of continuous diffusion models to discrete state spaces. NFDM learns a multivariate forward process from the data, ensuring that the forward process and generative trajectory are a good fit for language modeling. Our model substantially reduces the likelihood gap with autoregressive models of the same size, while achieving sample quality comparable to that of previous latent diffusion models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_16220 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Towards Latent Diffusion Suitable For Text Midavaine, Nesta Naesseth, Christian A. Bartosh, Grigory Computation and Language Machine Learning I.2.7 Language diffusion models aim to improve sampling speed and coherence over autoregressive LLMs. We introduce Neural Flow Diffusion Models for language generation, an extension of NFDM that enables the straightforward application of continuous diffusion models to discrete state spaces. NFDM learns a multivariate forward process from the data, ensuring that the forward process and generative trajectory are a good fit for language modeling. Our model substantially reduces the likelihood gap with autoregressive models of the same size, while achieving sample quality comparable to that of previous latent diffusion models. |
| title | Towards Latent Diffusion Suitable For Text |
| topic | Computation and Language Machine Learning I.2.7 |
| url | https://arxiv.org/abs/2601.16220 |