Research map / Probabilistic and scientific modeling / Variational and generative sampling
Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 variational rather than uniform dequantization (ablation) | Flow++ · CIFAR10 | bits/dim: 3.292 -> 3.165 | Helped |
| Use logistic mixture coupling rather than affine coupling (ablation) | Flow++ · CIFAR10 | bits/dim: 3.200 -> 3.165 | Helped |
| Add self-attention to the conditioning network (ablation) | Flow++ · CIFAR10 | bits/dim: 3.193 -> 3.165 | Helped |
| Use variational rather than uniform dequantization (ablation) | Flow++ · CIFAR10, 400 epochs | train-test gap (bits/dim): approximately 0.06 -> approximately 0.02 | Helped |
| Use Flow++ instead of Glow | Flow++ · CIFAR10 | bits/dim: 3.35 -> 3.08 | Helped |
| Use Flow++ instead of Glow | Flow++ · ImageNet 32x32 | bits/dim: 4.09 -> 3.86 | Helped |
| Use Flow++ instead of Glow | Flow++ · ImageNet 64x64 | bits/dim: 3.81 -> 3.69 | Helped |
Closest papers by meaning
- Analysis of Learning a Flow-based Generative Model from Limited Sample Complexity (ICLR 2024 · similarity 0.90)
- Normalizing Flows are Capable Generative Models (ICML 2025 · similarity 0.89)
- DiverseFlow: Sample-Efficient Diverse Mode Coverage in Flows (CVPR 2025 · similarity 0.89)
- Mean Flows for One-step Generative Modeling (NeurIPS 2025 · similarity 0.88)
- Align Your Flow: Scaling Continuous-Time Flow Map Distillation (NeurIPS 2025 · similarity 0.88)
- Faster Inference of Flow-Based Generative Models via Improved Data-Noise Coupling (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.