# Beyond the Loss Curve: Scaling Laws, Active Learning, and the Limits of Learning from Exact Posteriors

arXiv: https://arxiv.org/abs/2602.00315 · HTML: https://mlsanity.com/paper/2602.00315/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 31 references that are in the map.
- First posted on arXiv 30 Jan 2026.

## Builds on (most cited first, 24 of 24)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [A ConvNet for the 2020s](https://mlsanity.com/paper/2201.03545/) (2022)
- [Training Compute-Optimal Large Language Models](https://mlsanity.com/paper/2203.15556/) (2022)
- [Improved Techniques for Training GANs](https://mlsanity.com/paper/1606.03498/) (2016)
- [Benchmarking Neural Network Robustness to Common Corruptions and Perturbations](https://mlsanity.com/paper/1903.12261/) (2019)
- [Density estimation using Real NVP](https://mlsanity.com/paper/1605.08803/) (2016)
- [Searching for MobileNetV3](https://mlsanity.com/paper/1905.02244/) (2019)
- [Glow: Generative Flow with Invertible 1x1 Convolutions](https://mlsanity.com/paper/1807.03039/) (2018)
- [Normalizing Flows for Probabilistic Modeling and Inference](https://mlsanity.com/paper/1912.02762/) (2019)
- [StarGAN v2: Diverse Image Synthesis for Multiple Domains](https://mlsanity.com/paper/1912.01865/) (2019)
- [Scaling Vision Transformers](https://mlsanity.com/paper/2106.04560/) (2021)
- [Deep Variational Information Bottleneck](https://mlsanity.com/paper/1612.00410/) (2016)
- [Deep Bayesian Active Learning with Image Data](https://mlsanity.com/paper/1703.02910/) (2017)
- [BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning](https://mlsanity.com/paper/1906.08158/) (2019)
- [A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasets](https://mlsanity.com/paper/1707.08819/) (2017)
- [Normalizing Flows are Capable Generative Models](https://mlsanity.com/paper/2412.06329/) (2024)
- [Multimodal Datasets with Controllable Mutual Information](https://mlsanity.com/paper/2510.21686/) (2025)

## Built on by (most cited first, 1 of 1)

- [Beyond State-of-the-Art: Standardising Environmental Impact Metrics for AI Research](https://mlsanity.com/paper/2610.01116/) (2026)

## Abstract (arXiv)

How close are neural networks to the best they could possibly do? Standard benchmarks cannot answer this because they lack access to the true posterior p(y|x). We use class-conditional normalizing flows as oracles that make exact posteriors tractable on realistic images (AFHQ, ImageNet). This enables five lines of investigation. Scaling laws: Prediction error decomposes into irreducible aleatoric uncertainty and reducible epistemic error; the epistemic component follows a power law in dataset size, continuing to shrink even when total loss plateaus. Limits of learning: The aleatoric floor is exactly measurable, and architectures differ markedly in how they approach it: ResNets exhibit clean power-law scaling while Vision Transformers stall in low-data regimes. Soft labels: Oracle posteriors contain learnable structure beyond class labels: training with exact posteriors outperforms hard labels and yields near-perfect calibration. Distribution shift: The oracle computes exact KL divergence of controlled perturbations, revealing that shift type matters more than shift magnitude: class imbalance barely affects accuracy at divergence values where input noise causes catastrophic degradation. Active learning: Exact epistemic uncertainty distinguishes genuinely informative samples from inherently ambiguous ones, improving sample efficiency. Our framework reveals that standard metrics hide ongoing learning, mask architectural differences, and cannot diagnose the nature of distribution shift.