Search papers, labs, and topics across Lattice.
This paper introduces DiSE, a self-evaluation method for diffusion language models (dLLMs) that quantifies confidence by computing the probability of regenerating the entire generated sequence given the full context. DiSE enables efficient quality assessment by leveraging token regeneration probabilities for likelihood estimation and uncertainty quantification in dLLMs. Experiments demonstrate DiSE's effectiveness in likelihood evaluation, uncertainty quantification, and flexible-length generation, showing positive correlation with semantic coherence and answer accuracy.
Diffusion language models can now efficiently self-evaluate their output quality by regenerating their own sequences, enabling more reliable uncertainty quantification and flexible-length generation.
Diffusion large language models (dLLMs) have recently attracted significant attention for their ability to enhance diversity, controllability, and parallelism. However, their non-sequential, bidirectionally masked generation makes quality assessment difficult, underscoring the need for effective self-evaluation. In this work, we propose DiSE, a simple yet effective self-evaluation confidence quantification method for dLLMs. DiSE quantifies confidence by computing the probability of regenerating the tokens in the entire generated sequence, given the full context. This method enables more efficient and reliable quality assessment by leveraging token regeneration probabilities, facilitating both likelihood estimation and robust uncertainty quantification. Building upon DiSE, we further introduce a flexible-length generation framework, which adaptively controls the sequence length based on the model's self-assessment of its own output. We analyze and validate the feasibility of DiSE from the perspective of dLLM generalization, and empirically demonstrate that DiSE is positively correlated with both semantic coherence and answer accuracy. Extensive experiments on likelihood evaluation, uncertainty quantification, and flexible-length generation further confirm the effectiveness of the proposed DiSE.
