- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 3
Read review
Paper summary
The paper studies how source–data SNR and architectural compression affect prediction-target preferences in flow matching. It proposes a binary data/noise switch and a continuous Gaussian-based parameterization with a learned residual, both using covariance information to adapt across directions and time. The paper provides Gaussian analyses and evaluates the methods on synthetic data and image generation.
Critical strengths
-
Informative analysis of prediction targets. The controlled experiments help distinguish the effects of ambient dimension and source–data SNR. The Gaussian analysis gives testable predictions about how parameterization interacts with compression, and the bottleneck experiments provide useful support for these predictions.
-
Promising image-generation results. The hybrids show robust performance across source scales on CelebA and reach low FID in fewer optimization steps on Bedroom. The convergence gains are observed with both U-Net and ViT models.
Critical weaknesses
- Limited evaluation at scale. Related work such as JiT (Li and He, 2025) and k-Diff (Jin and Wang, 2026) reports quantitative results on ImageNet. The experiments here remain smaller in scale, and the higher-resolution results are mainly qualitative. It is unclear whether the reported gains persist on larger and more diverse datasets. Actual training-time measurements would also help assess the efficiency claims.
- Unclear added benefit of directional adaptation. The image-generation comparisons leave it unclear how much is gained from adapting individual directions beyond adapting to global SNR. The uniform-switch baseline uses a fixed switching time, so it does not test whether an SNR-dependent rule shared across all directions could provide similar gains.
Questions for authors
-
Could you provide quantitative results on a larger image-generation benchmark, such as ImageNet?
-
Could you compare the binary hybrid with a baseline that switches all directions together at a time determined by global SNR? For the continuous hybrid, could you compare with a version that uses one time-dependent coefficient for the analytical input term, shared across directions and based on average data variance and source-noise scale?
-
How do the hybrids compare with fixed-target baselines in wall-clock time to reach a comparable FID, including preprocessing and spectral operations?
I would consider raising my rating if the response provides stronger evidence at scale and clarifies the benefit of directional adaptation.
2 comments
Thank you for the response and additional experiments. These additions largely address my main concerns, and I will raise my rating from 2 to 3. Please include the additional results and implementation details in the revised paper.
We thank the reviewer for their positive assessment and constructive suggestions.
W1 & Q1. Evaluation at scale
We extended our evaluation to ImageNet-100 at 256×256. The results are very positive: at matched training stages, the continuous hybrid substantially outperforms x1-prediction and converges markedly faster. See Point 1 of the general response for quantitative results and implementation details.
W2 & Q2. Benefit of direction-dependent adaptation
We thank the reviewer for suggesting this ablation. We added global binary and continuous hybrids that retain SNR-dependent adaptation but apply the same rule to every covariance direction. See Point 3 of the general response for the definitions and quantitative results.
The result is particularly clear: both global hybrids perform worse than x1-prediction, whereas both spectral hybrids substantially outperform it. This directly shows that adapting only to the global SNR is insufficient, and that direction-dependent adaptation is essential to the observed gains.
Q3. Wall-clock training efficiency
The preprocessing cost is incurred only once: the spectral statistics are computed, cached, and reused across runs. For CelebA-64, the full PCA requires about 0.6 GB of memory and takes 4 minutes to compute on a V100 GPU.
During training, the additional cost mainly consists of matrix-vector products, which are negligible compared with the network forward and backward passes. We now report FID as a function of wall-clock time on LSUN-Bedroom-64 (Figure 6 b). The resulting curves preserve the gains observed versus optimization steps, showing that the spectral hybrids reach a given FID substantially earlier than the best fixed-target baseline.
At higher resolution, we avoid the O(D2) memory cost of full-image PCA by using scalable spectral representations: a fixed 2D DCT basis for AFHQ-256 and patch-level PCA for ImageNet-100 (see common answer, Point 1).