Salvato in:
Dettagli Bibliografici
Autori principali: Fang, Zhengyu, Jiang, Zhimeng, Chen, Huiyuan, Zhang, Xiaoge, Li, Tianyi, Tang, Kaiyu, Li, Xiao, Li, Jing
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2604.13413
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918447507570688
author Fang, Zhengyu
Jiang, Zhimeng
Chen, Huiyuan
Zhang, Xiaoge
Li, Tianyi
Tang, Kaiyu
Li, Xiao
Li, Jing
author_facet Fang, Zhengyu
Jiang, Zhimeng
Chen, Huiyuan
Zhang, Xiaoge
Li, Tianyi
Tang, Kaiyu
Li, Xiao
Li, Jing
contents Diffusion language models (DLMs) have emerged as a promising paradigm for large language models (LLMs), yet the non-deterministic behavior of DLMs remains poorly understood. The existing non-determinism evaluations for LLMs predominantly rely on dataset-level metrics under fixed inference configurations, providing limited insight into how model behavior varies across runs and evaluation conditions. In this work, we show that dataset-level metrics systematically attenuate non-determinism in diffusion language models by aggregating sample-level prediction quality across different runs. As a result, configurations with similar aggregate performance can exhibit substantially different behaviors on individual inputs, leaving fine-grained instability and distinct error patterns uncharacterized. To address this limitation, we conduct a fine-grained evaluation of non-determinism based on sample-level prediction differences across a range of model-related factors-including guidance scale, diffusion steps, and Monte Carlo sampling-as well as system-related factors such as batch size, hardware, and numerical precision. Our analysis reveals that non-determinism in DLMs is pervasive and structured, with code generation exhibiting markedly higher sensitivity to factor-level choices than question answering. To attribute sources of non-determinism evaluation, we introduce Factor Variance Attribution (FVA), a cross-factor analysis metric that decomposes observed non-determinism into variance attributable to different evaluation factor settings. Our findings highlight the need for fine-grained, factor-aware evaluation to enable reliable non-determinism assessment of diffusion language models.
format Preprint
id arxiv_https___arxiv_org_abs_2604_13413
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Dataset-Level Metrics Attenuate Non-Determinism: A Fine-Grained Non-Determinism Evaluation in Diffusion Language Models
Fang, Zhengyu
Jiang, Zhimeng
Chen, Huiyuan
Zhang, Xiaoge
Li, Tianyi
Tang, Kaiyu
Li, Xiao
Li, Jing
Machine Learning
Diffusion language models (DLMs) have emerged as a promising paradigm for large language models (LLMs), yet the non-deterministic behavior of DLMs remains poorly understood. The existing non-determinism evaluations for LLMs predominantly rely on dataset-level metrics under fixed inference configurations, providing limited insight into how model behavior varies across runs and evaluation conditions. In this work, we show that dataset-level metrics systematically attenuate non-determinism in diffusion language models by aggregating sample-level prediction quality across different runs. As a result, configurations with similar aggregate performance can exhibit substantially different behaviors on individual inputs, leaving fine-grained instability and distinct error patterns uncharacterized. To address this limitation, we conduct a fine-grained evaluation of non-determinism based on sample-level prediction differences across a range of model-related factors-including guidance scale, diffusion steps, and Monte Carlo sampling-as well as system-related factors such as batch size, hardware, and numerical precision. Our analysis reveals that non-determinism in DLMs is pervasive and structured, with code generation exhibiting markedly higher sensitivity to factor-level choices than question answering. To attribute sources of non-determinism evaluation, we introduce Factor Variance Attribution (FVA), a cross-factor analysis metric that decomposes observed non-determinism into variance attributable to different evaluation factor settings. Our findings highlight the need for fine-grained, factor-aware evaluation to enable reliable non-determinism assessment of diffusion language models.
title Dataset-Level Metrics Attenuate Non-Determinism: A Fine-Grained Non-Determinism Evaluation in Diffusion Language Models
topic Machine Learning
url https://arxiv.org/abs/2604.13413