Research map / Architectures and efficiency / Quantization
Quantized Visual Geometry Grounded Transformer
Weilun Feng, Haotong Qin, Mingqiang Wu, Chuanguang Yang, Yuqi Li, Xiangqi Li, Zhulin An, Libo Huang, Yulun Zhang, Michele Magno, Yongjun Xu
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 |
|---|---|---|---|
| add rotation + scale smoothing (DSFQ) to naive W4A4 quantization (ablation) | VGGT-1B W4A4 · CO3Dv2 10 frames | AUC@3: 9.7 -> 43.6 | Improved |
| add post-local scaling on top of rotation (ablation) | VGGT-1B W4A4 · CO3Dv2 10 frames | AUC@3: 32.5 -> 43.6 | Improved |
| add Hadamard rotation on top of scaling (ablation) | VGGT-1B W4A4 · CO3Dv2 10 frames | AUC@3: 21.2 -> 43.6 | Improved |
| rotate first then smooth (instead of smooth then rotate) (ablation) | VGGT-1B W4A4 · CO3Dv2 10 frames | AUC@3: 42.9 -> 43.6 | Improved |
| QuantVGGT vs QuaRot at W4A4 | VGGT-1B · CO3Dv2 20 frames | AUC@30: 81.6 -> 88.2 | Improved |
| QuantVGGT vs QuaRot at W4A4 | VGGT-1B · DTU point map | Acc. mean (lower better): 1.593 -> 1.282 | Improved |
| QuantVGGT at W8A8 | VGGT-1B · DTU point map | Acc. mean (lower better): 1.185 -> 1.182 | No clear effect |
| noise-filtered diverse sampling vs random calibration sampling (ablation) | QuantVGGT · CO3Dv2 W4A4 | AUC variance across seeds | Improved |
| apply SmoothQuant at W4A4 | VGGT-1B · CO3Dv2 10 frames | AUC@30: vs RTN 77.7 -> 75.4 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- QuaRot — QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs (NeurIPS 2024) · better on 8 of 8 shared results
- SmoothQuant — SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models (ICML 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
- Not All Tasks Quantize Equally: Fisher-Guided Quantization for Visual Geometry Transformer (NeurIPS 2026 · similarity 0.91)
- RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer (NeurIPS 2026 · similarity 0.89)
- VGGT: Visual Geometry Grounded Transformer (CVPR 2025 · similarity 0.89)
- Outlier-aware Slicing for Post-Training Quantization in Vision Transformer (ICML 2024 · similarity 0.88)
- GradQ-ViT: Robust and Efficient Gradient Quantization for Vision Transformers (AAAI 2025 · similarity 0.88)
- ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generation (ICLR 2025 · similarity 0.88)
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.
