Research map / Video / Video generation
AdapTok: Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space
Yan Li, Changyao Tian, Renqiu Xia, Ning Liao, Weiwei Guo, Hongsheng Li, Jifeng Dai, Hao Li, Xue Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use AdapTok adaptive tokenizer instead of non-adaptive CausalTok | AdapTok · UCF-101 reconstruction | rFVD: 37 -> 36 (1,024 tokens); 60 with 512 tokens | Improved |
| use AdapTok tokens instead of CausalTok tokens for autoregressive generation | AR generator · UCF-101 class-conditional generation | gFVD: 80 -> 67 | Improved |
| combine block-mask sampler training with adaptive scorer inference (ablation) | AdapTok · UCF-101 reconstruction | rFVD: best rFVD 36.36 | Improved |
| use ILP allocation instead of fixed, BiThr or BiDelta allocation (ablation) | AdapTok · UCF-101 reconstruction | FVD, PSNR, LPIPS | Improved |
| use perceptual loss as the scoring metric instead of SSIM, PSNR or MSE (ablation) | adaptive scorer · UCF-101 reconstruction | rFVD | Mixed |
| scale model from AdapTok-S to AdapTok-L (ablation) | AdapTok · UCF-101 reconstruction | rFVD: 87.32 -> 36.36 | Improved |
| scale model from AdapTok-L to AdapTok-XL (ablation) | AdapTok · UCF-101 reconstruction | rFVD: 36.36 -> 32.43 | Improved |
| use scorer with IPAL instead of ElasticTok threshold allocation | AdapTok · inference latency | latency: 11x faster | Improved |
| increase ILP mini-batch size from 8 to 1024 (ablation) | IPAL · UCF-101 reconstruction, avg 1024 tokens | FVD: 36.55 -> 37.01 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Phenaki — Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions (ICLR 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ElasticTok: Adaptive Tokenization for Image and Video (ICLR 2025 · similarity 0.88)
- LARP: Tokenizing Videos with a Learned Autoregressive Generative Prior (ICLR 2025 · similarity 0.87)
- Efficient Long Video Tokenization via Coordinate-based Patch Reconstruction (CVPR 2025 · similarity 0.87)
- Extending Video Masked Autoencoders to 128 frames (NeurIPS 2024 · similarity 0.86)
- Training-Free Efficient Video Generation via Dynamic Token Carving (NeurIPS 2025 · similarity 0.86)
- DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
