LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914580207239168 |
|---|---|
| author | Chen, Yukang Wang, Luozhou Huang, Wei Yang, Shuai Zhang, Bohan Xiao, Yicheng Chu, Ruihang Mao, Weian Hu, Qixin Liu, Shaoteng Zhao, Yuyang Mao, Huizi Chen, Ying-Cong Xie, Enze Qi, Xiaojuan Han, Song |
| author_facet | Chen, Yukang Wang, Luozhou Huang, Wei Yang, Shuai Zhang, Bohan Xiao, Yicheng Chu, Ruihang Mao, Weian Hu, Qixin Liu, Shaoteng Zhao, Yuyang Mao, Huizi Chen, Ying-Cong Xie, Enze Qi, Xiaojuan Han, Song |
| contents | We present LongLive-2.0, an NVFP4-based parallel infrastructure throughout the full training and inference workflow of long video generation, addressing speed and memory bottlenecks. For training, we introduce sequence-parallel autoregressive (AR) training, instantiated as Balanced SP, which co-designs the efficient teacher-forcing layout with SP execution by pairing clean-history and noisy-target temporal chunks on each rank, enabling a natural teacher-forcing mask with SP-aware chunked VAE encoding. Combined with NVFP4 precision, it reduces GPU memory cost and accelerates GEMM computation during training, the proportion of which increases as video length grows. Moreover, we show that a high-quality infrastructure and dataset enable a remarkably clean training pipeline. Unlike existing Self-Forcing series methods that rely on ODE initialization and subsequent distribution matching distillation (DMD), LongLive-2.0 directly tunes a diffusion model into a long, multi-shot, interactive auto-regressive (AR) diffusion model. It can be further converted to real-time generation (4 to 2 denoising steps) with standalone LoRA weights. For inference on Blackwell GPUs, we enable W4A4 NVFP4 inference, quantize KV cache into NVFP4 for memory savings, and boost end-to-end throughput with asynchronous streaming VAE decoding. On non-Blackwell GPU architectures, we deploy SP inference to match the speed on Blackwell GPUs, while the quantized KV cache can lower inter-GPU communication of SP. Experiments show up to 2.15x speedup in training, and 1.84x in inference. LongLive-2.0-5B achieves 45.7 FPS inference while attaining strong performance on benchmarks. To our knowledge, LongLive-2.0 is the first NVFP4 training and inference system for long video generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_18739 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation Chen, Yukang Wang, Luozhou Huang, Wei Yang, Shuai Zhang, Bohan Xiao, Yicheng Chu, Ruihang Mao, Weian Hu, Qixin Liu, Shaoteng Zhao, Yuyang Mao, Huizi Chen, Ying-Cong Xie, Enze Qi, Xiaojuan Han, Song Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing We present LongLive-2.0, an NVFP4-based parallel infrastructure throughout the full training and inference workflow of long video generation, addressing speed and memory bottlenecks. For training, we introduce sequence-parallel autoregressive (AR) training, instantiated as Balanced SP, which co-designs the efficient teacher-forcing layout with SP execution by pairing clean-history and noisy-target temporal chunks on each rank, enabling a natural teacher-forcing mask with SP-aware chunked VAE encoding. Combined with NVFP4 precision, it reduces GPU memory cost and accelerates GEMM computation during training, the proportion of which increases as video length grows. Moreover, we show that a high-quality infrastructure and dataset enable a remarkably clean training pipeline. Unlike existing Self-Forcing series methods that rely on ODE initialization and subsequent distribution matching distillation (DMD), LongLive-2.0 directly tunes a diffusion model into a long, multi-shot, interactive auto-regressive (AR) diffusion model. It can be further converted to real-time generation (4 to 2 denoising steps) with standalone LoRA weights. For inference on Blackwell GPUs, we enable W4A4 NVFP4 inference, quantize KV cache into NVFP4 for memory savings, and boost end-to-end throughput with asynchronous streaming VAE decoding. On non-Blackwell GPU architectures, we deploy SP inference to match the speed on Blackwell GPUs, while the quantized KV cache can lower inter-GPU communication of SP. Experiments show up to 2.15x speedup in training, and 1.84x in inference. LongLive-2.0-5B achieves 45.7 FPS inference while attaining strong performance on benchmarks. To our knowledge, LongLive-2.0 is the first NVFP4 training and inference system for long video generation. |
| title | LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation |
| topic | Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2605.18739 |