# Unconditional Image Modeling on ImageNet 32x32

Source: [Unconditional Image Modeling on ImageNet 32x32](<https://www.sota2.com/research/sota/unconditional-image-modeling-on-imagenet-32x32>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:37:51.670Z

Evaluation of unconditional image modeling performance (density estimation or likelihood) on the ImageNet dataset rendered at 32x32 resolution, typically measured using bits/dim.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Unconditional image modeling |
| Dataset (including any stated split) | ImageNet 32x32 |
| Metric | Bits/Dim |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 8 of 8 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Bits/Dim](<https://www.sota2.com/research/sota/unconditional-image-modeling-on-imagenet-32x32.md?metric=Bits%2FDim>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Bits/Dim | Result date | Report paper |
| --- | --- | --- | --- |
| Image Transformer | 3.77 | Not recorded | 68fa65be1e209131cea0c11d |
| PixelSNAIL | 3.8 | Not recorded | 68fa65be1e209131cea0c11d |
| Gated PixelCNN | 3.83 | Not recorded | 68fa65be1e209131cea0c11d |
| Flow++ | 3.86 | Not recorded | 68fa65be1e209131cea0c11d |
| PixelRNN | 3.86 | Not recorded | 68fa65be1e209131cea0c11d |
| Multiscale PixelCNN | 3.95 | Not recorded | 68fa65be1e209131cea0c11d |
| Glow | 4.09 | Not recorded | 68fa65be1e209131cea0c11d |
| RealNVP | 4.28 | Not recorded | 68fa65be1e209131cea0c11d |

### Image Transformer

Bits/Dim: 3.77. Result ID: 698cc5518b517eb4331c9f85.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64831e209131cea050c0.md>)

Recorded evaluation configuration: {"Model family":"Autoregressive"}. This record does not assert independent reproduction.

### PixelSNAIL

Bits/Dim: 3.8. Result ID: 698cc5518b517eb4331c9f86.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63e81e209131cea015ac.md>)

Recorded evaluation configuration: {"Model family":"Autoregressive"}. This record does not assert independent reproduction.

### Gated PixelCNN

Bits/Dim: 3.83. Result ID: 698cc5518b517eb4331c9f84.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa629e1e209131ce9fa3c3.md>)

Recorded evaluation configuration: {"Model family":"Autoregressive"}. This record does not assert independent reproduction.

### Flow++

Bits/Dim: 3.86. Result ID: 698cc5518b517eb4331c9f81.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Implementation](<https://github.com/aravindsrinivas/flowpp>)

Recorded evaluation configuration: {"Model family":"Non-autoregressive"}. This record does not assert independent reproduction.

### PixelRNN

Bits/Dim: 3.86. Result ID: 698cc5518b517eb4331c9f83.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62b81e209131ce9fad7e.md>)

Recorded evaluation configuration: {"Model family":"Autoregressive"}. This record does not assert independent reproduction.

### Multiscale PixelCNN

Bits/Dim: 3.95. Result ID: 698cc5518b517eb4331c9f82.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa632f1e209131ce9fd244.md>)

Recorded evaluation configuration: {"Model family":"Autoregressive"}. This record does not assert independent reproduction.

### Glow

Bits/Dim: 4.09. Result ID: 698cc5518b517eb4331c9f80.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa649e1e209131cea05a05.md>)

[Implementation](<https://github.com/openai/glow>)

Recorded evaluation configuration: {"Model family":"Non-autoregressive"}. This record does not assert independent reproduction.

### RealNVP

Bits/Dim: 4.28. Result ID: 698cc5518b517eb4331c9f7f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c11d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63351e209131ce9fd6b3.md>)

Recorded evaluation configuration: {"Model family":"Non-autoregressive"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.