BARD：通过高效渐进性块合并和分阶段蒸馏桥接自回归与扩散视觉-语言模型
BARD: Bridging AutoRegressive and Diffusion Vision-Language Models Via Highly Efficient Progressive Block Merging and Stage-Wise Distillation
AI总结：
BARD通过高效渐进块合并与分阶段蒸馏将预训练自回归视觉-语言模型转换为解码高效的扩散视觉-语言模型，实现在大块规模下性能恢复与解码效率提升。
摘要：
自回归视觉-语言模型（VLMs）具备强大的多模态能力，但其逐token解码方式存在根本性的推理瓶颈。扩散VLMs提供了一种更并行的解码范式，但直接将预训练的自回归VLM转换为大块扩散VLM（dVLM）往往导致显著的质量下降。本文提出BARD，一种简单有效的桥接框架，将预训练的自回归VLM转换为同架构、解码高效的dVLM。我们的方法结合了渐进式监督块合并，逐步扩大解码块大小，与分阶段的内在dVLM蒸馏，从固定的小块扩散锚点进行蒸馏以恢复在较大块中丢失的性能。我们进一步引入混合噪声调度器以提高去噪过程中的鲁棒性和token修正，并采用内存友好的训练以实现对长多模态序列的高效训练。关键的实证发现是，直接自回归到扩散的蒸馏对齐不佳，甚至可能损害性能，而扩散范式内的蒸馏始终有效。实验结果表明，BARD-VL在$\leq$4.4M数据下将强多模态能力从Qwen3-VL转移到大块dVLM。值得注意的是，BARD-VL在我们的评估套件上，在4B和8B规模中建立了新的SOTA。同时，BARD-VL在源模型上实现了高达3$\times$的解码吞吐量提升。代码可在https://github.com/fudan-generative-vision/Bard-VL获取。
英文摘要：
Autoregressive vision-language models (VLMs) deliver strong multimodal capability, but their token-by-token decoding imposes a fundamental inference bottleneck. Diffusion VLMs offer a more parallel decoding paradigm, yet directly converting a pretrained autoregressive VLM into a large-block diffusion VLM (dVLM) often leads to substantial quality degradation. In this work, we present BARD, a simple and effective bridging framework that converts a pretrained autoregressive VLM into a same-architecture, decoding-efficient dVLM. Our approach combines progressive supervised block merging, which gradually enlarges the decoding block size, with stage-wise intra-dVLM distillation from a fixed small-block diffusion anchor to recover performance lost at larger blocks. We further incorporate a mixed noise scheduler to improve robustness and token revision during denoising, and memory-friendly training to enable efficient training on long multimodal sequences. A key empirical finding is that direct autoregressive-to-diffusion distillation is poorly aligned and can even hurt performance, whereas distillation within the diffusion regime is consistently effective. Experimental results show that, with $\leq$ 4.4M data, BARD-VL transfers strong multimodal capability from Qwen3-VL to a large-block dVLM. Remarkably, BARD-VL establishes a new SOTA among comparable-scale open dVLMs on our evaluation suite at both 4B and 8B scales. At the same time, BARD-VL achieves up to 3$\times$ decoding throughput speedup compared to the source model. Code is available at https://github.com/fudan-generative-vision/Bard-VL.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
