研究人员开发了一种新颖的方法，通过迭代改进其码本（codebooks）来增强自回归图像生成模型的安全性。该方法利用模型自身的理解来识别和消除有害的图像-文本映射，然后使用安全示例对码本进行微调，以保持生成质量。另外一项研究引入了ScalingAR，一个用于测试时缩放的框架，旨在通过使用令牌熵（token entropy）作为置信度信号来提高自回归图像生成的效率和鲁棒性，从而在无需早期解码或外部奖励模型的情况下获得显著的性能提升。
AI
arXiv:2606.27978v1 Announce Type: cross Abstract: Pixel-space continuous-token autoregressive (AR) generation directly models images as sequences of raw pixel patches, avoiding discrete tokenization or a separately pretrained tokenizer. However, it faces coupled challenges: high-…
Pixel-space continuous-token autoregressive (AR) generation directly models images as sequences of raw pixel patches, avoiding discrete tokenization or a separately pretrained tokenizer. However, it faces coupled challenges: high-dimensional patch generation causes large single-s…
arXiv cs.AI
TIER_1English(EN)·Yunqi Xue, Zhijiang Li, Philip Torr, Jindong Gu·
arXiv:2606.27147v1 Announce Type: cross Abstract: Unlike diffusion-based models that operate in continuous latent spaces, autoregressive unified multimodal models produce images by sequentially predicting discretized visual tokens. These tokens are derived from a codebook that ma…
Parallel Rollout Approximation (PRA) addresses limitations in pixel-space autoregressive image generation by using low-dimensional intermediate states and parallel training to improve quality and efficiency.
Unlike diffusion-based models that operate in continuous latent spaces, autoregressive unified multimodal models produce images by sequentially predicting discretized visual tokens. These tokens are derived from a codebook that maps embeddings to quantized visual patterns. The la…
Unlike diffusion-based models that operate in continuous latent spaces, autoregressive unified multimodal models produce images by sequentially predicting discretized visual tokens. These tokens are derived from a codebook that maps embeddings to quantized visual patterns. The la…
arXiv:2509.26376v3 Announce Type: replace Abstract: Test-time strategies have shown remarkable success in improving large language models, but their application to next-token prediction (NTP) autoregressive (AR) image generation remains largely underexplored. Existing test-time s…
