DreamReasoner-8B: Block-Size Curriculum Learning for Diffusion Reasoning Models
This work addresses the scalability of block diffusion models for long reasoning tasks, which is a known bottleneck for efficient inference in language models.
DreamReasoner-8B introduces block-size curriculum learning to train block diffusion language models for long chain-of-thought reasoning, achieving competitive results with leading autoregressive models like Qwen3-8B on math and code benchmarks.
Block diffusion language models accelerate decoding through parallel block-wise denoising, yet whether they can be reliably scaled for long chain-of-thought (CoT) reasoning remains unresolved. To this end, we develop DreamReasoner-8B, an open-source block diffusion reasoning model, and conduct a systematic study of how training and inference block sizes affect long-CoT reasoning. Our analysis reveals a stark performance disparity: training with large block sizes yields remarkably poor reasoning, whereas small block sizes preserve effective reasoning. To bridge this granularity gap, we propose block-size curriculum learning, which gradually transitions training from fine-grained to coarse-grained block sizes, thereby overcoming this limitation and enabling strong reasoning performance that generalizes across diverse inference block sizes. On mathematical and code reasoning benchmarks, DreamReasoner-8B achieves results competitive with leading open autoregressive models such as Qwen3-8B. This work establishes a practical foundation for efficient, reasoning-capable diffusion language models. We release our model at https://github.com/DreamLM/DreamReasoner.