Research map / Image generation / Super-resolution
Learning Implicit Generative Models by Matching Perceptual Features
Cicero Nogueira dos Santos, Youssef Mroueh, Inkit Padhi, Pierre Dognin
This paper tried a common change: Larger batch size
increase minibatch size from 64 to 512 with ADAM moving average · CelebA · image quality (qualitative) No clear effect
What 785 accepted papers report for this change
- 469 improved
- 107 no clear effect
- 48 mixed
- 161 hurt
- Predicting the Emergence of Induction Heads in Language Model Pretraining ICML 2026 · increase batch size (fixed context 1024) · Natural data (CC100) · induction head emergence point (updates): rho=-1.0, p<0.001Improved
- When RL Meets Adaptive Speculative Training: A Unified Training-Serving System ICML 2026 · Increase batch size from 1 to 32 · MiniMax M2.1 FP8 · speedup: 1.57x -> 1.25xHurt
- Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking ICML 2026 · increase per-device batch size to 10 using freed memory · Llama 3-8B, 128K · throughput (TPS): 2464.10 (Ulysses) -> 2526.77 (UPipe)Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 ImageNet VGG19 classifier features instead of VGG19 encoder features (ablation) | GFMN · CIFAR10 | Inception Score: 4.95 -> 7.88 | Improved |
| use ImageNet VGG19 classifier features instead of VGG19 encoder features (ablation) | GFMN · CIFAR10 | FID (50K): 57.2 -> 20.8 | Improved |
| use VGG19 + Resnet18 features together instead of VGG19 alone (ablation) | GFMN · CIFAR10 | Inception Score: 7.88 -> 8.08 | Improved |
| use VGG19 + Resnet18 features together instead of VGG19 alone (ablation) | GFMN · CIFAR10 | FID (50K): 20.8 -> 20.9 | No clear effect |
| replace DCGAN generator with Resnet generator (ablation) | GFMN VGG19+Resnet18 · CIFAR10 | Inception Score: 8.08 -> 8.27 | Improved |
| replace DCGAN generator with Resnet generator (ablation) | GFMN VGG19+Resnet18 · CIFAR10 | FID (50K): 20.9 -> 13.5 | Improved |
| increase number of VGG19 layers used for matching from 1 to 16 (ablation) | GFMN VGG19 · CIFAR10 | Inception Score: 4.68 -> 7.88 | Improved |
| increase number of VGG19 layers from 11 to 13 (ablation) | GFMN VGG19 · CIFAR10 | Inception Score: 7.72 -> 7.49 | No clear effect |
| add covariance matching to mean matching (VGG19 features) (ablation) | GFMN · CIFAR10 | Inception Score: 7.42 -> 7.88 | Improved |
| add covariance matching to mean matching (DCGAN encoder features) (ablation) | GFMN · CIFAR10 | FID (50K): 92.5 -> 78.3 | Improved |
| add covariance matching to mean matching (Resnet18 features) (ablation) | GFMN · CIFAR10 | Inception Score: 7.03 -> 7.92 | Improved |
| replace simple moving average with ADAM moving average at minibatch size 64 (ablation) | GFMN · CelebA | image quality (qualitative) | Improved |
Closest papers by meaning
- Learning Implicit Generative Models with the Method of Learned Moments (ICML 2018 · similarity 0.88)
- Conditional Generative Moment-Matching Networks (NeurIPS 2016 · similarity 0.86)
- Inductive Moment Matching (ICML 2025 · similarity 0.83)
- Feature Unlearning for Pre-trained GANs and VAEs (AAAI 2024 · similarity 0.83)
- Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator (ICML 2025 · similarity 0.83)
- Exact Gaussian Moment Matching for Residual Networks: a Second-Order Method (NeurIPS 2026 · similarity 0.83)
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.
