DiffER: Diffusion Entity-Relation Modeling for Reversal Curse in Diffusion Large Language Models
The "reversal curse" refers to the phenomenon where large language models (LLMs) exhibit predominantly unidirectional behavior when processing logically bidirectional relationships. Prior work attributed this to autoregressive training -- predicting the next token inherently favors left-to-right information flow over genuine bidirectional knowledge associations. However, we observe that Diffusion LLMs (DLLMs), despite being trained bidirectionally, also suffer from the reversal curse. To investigate the root causes, we conduct systematic experiments on DLLMs and identify three key reasons: 1) entity fragmentation during training, 2) data asymmetry, and 3) missing entity relations. Motivated by the analysis of these reasons, we propose Diffusion Entity-Relation Modeling (DiffER), which addresses the reversal curse through entity-aware training and balanced data construction. Specifically, DiffER introduces whole-entity masking, which mitigates entity fragmentation by predicting complete entities in a single step. DiffER further employs distribution-symmetric and relation-enhanced data construction strategies to alleviate data asymmetry and missing relations. Extensive experiments demonstrate that DiffER effectively alleviates the reversal curse in Diffusion LLMs, offering new perspectives for future research.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Non-Denoising Forward-Time Diffusions
The scope of this paper is generative modeling through diffusion processes. An approach falling within this paradigm is the work of Song et al. (2021), which relies on a time-reversal argument to construct a diffusion pr…
DenoisingThe Illusion of Latent Generalization: Bi-directionality and the Reversal Curse
The reversal curse describes a failure of autoregressive language models to retrieve a fact in reverse order (e.g., training on ``$A > B$'' but failing on ``$B < A$''). Recent work shows that objectives with bidirectiona…
Memorization, Emergence, and Explaining Reversal Failures: A Controlled Study of Relational Semantics in LLMs
Autoregressive LLMs perform well on relational tasks that require linking entities via relational words (e.g., father/son, friend), but it is unclear whether they learn the logical semantics of such relations (e.g., symm…
Denoising Diffusion Samplers
Denoising diffusion models are a popular class of generative models providing state-of-the-art results in many domains. One adds gradually noise to data using a diffusion to transform the data distribution into a Gaussia…
DenoisingAn effective potential for generative modelling with active matter
Score-based diffusion models generate samples from a complex underlying data distribution by time-reversal of a diffusion process and represent the state-of-the-art in many generative AI applications. Here, I show how a …