dQwen3.5: Hybrid-Attention Diffusion Language Models
Anton Xue, Litu Rout, Aditya Akella, Adam Klivans, Sujay Sanghavi, Sanjay Shakkottai
Abstract
Adapting a pretrained autoregressive (AR) model is a cost-efficient route to a diffusion language model (DLM). While nearly all such adaptations start from a full-attention transformer, AR modeling has shifted toward hybrid architectures that interleave attention and RNN layers. This creates an obstacle for adaptation: unlike attention, RNNs are structurally causal and nontrivial to bidirectionalize. Despite this mismatch, we investigate whether such backbones can become effective DLMs by adapting Qwen3.5 at 0.8B, 2B, 4B, and 9B scales, yielding the dQwen3.5 family. We find that hybrid backbones can be efficient starting points for adaptation: against a full-attention control, the hybrid reaches a given training loss in about half the tokens. Across scales, dQwen3.5 resembles full-attention DLMs in any-order decoding behavior and performs strongly under parallel decoding.
Create a lesson
Related papers
RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning
Yan Yu, Zhengxi Lu, Yizhou Liu et al.
Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations
Sarah Wyer, Sue Black, Noura Al Moubayed
On-Demand Attention: Language Models Know When to Recall
Haibo Feng, Ruiqi Liang, Hanyang Peng et al.
Summarization Bias: The Directional Collapse of Objective Projection into Told-Mode Labels in Large Language Models --- A Conceptual Framework and Registered Test Protocol
Levent Bulut
HerHealthEval: Evaluating Multilingual and Register-Sensitive Understanding of Women's Health Communication
Hassan Saeed Hassan Albattra, Mazen Mohammed Bahgat, Rahatara Ferdousi et al.
UniPolicy: Unified Objective-Specific Policies for Generative Search Advertising
Kun Yao, Yuhang Zhou, Yichi Zhang et al.