# Density Estimation on ImageNet 64 x 64

Source: [Density Estimation on ImageNet 64 x 64](<https://www.sota2.com/research/sota/density-estimation-on-imagenet-64-x-64>)

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

Data updated: 2026-02-26T05:53:17.508Z

Evaluation of Density Estimation performance on the ImageNet dataset downsampled to 64x64 resolution, typically measured using Negative Log-Likelihood (NLL) in bits per dimension.

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 | Density Estimation |
| Dataset (including any stated split) | ImageNet 64 x 64 |
| Metric | NLL (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

- [NLL (bits/dim)](<https://www.sota2.com/research/sota/density-estimation-on-imagenet-64-x-64.md?metric=NLL%20(bits%2Fdim)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | NLL (bits/dim) | Result date | Report paper |
| --- | --- | --- | --- |
| Efficient-VDVAE | 3.3 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| DenseFlow | 3.35 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| VDM | 3.4 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| Sparse Transformer | 3.44 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| SPN | 3.52 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| VDVAE | 3.52 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| Flow++ | 3.69 | Not recorded | 68fa6c6a1e209131cea33fb1 |
| GLOW | 3.81 | Not recorded | 68fa6c6a1e209131cea33fb1 |

### Efficient-VDVAE

NLL (bits/dim): 3.3. Result ID: 698f13533a6c314849f5840f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Implementation](<https://github.com/Rayhane-mamah/Efficient-VDVAE>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DenseFlow

NLL (bits/dim): 3.35. Result ID: 698f13533a6c314849f5840a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b271e209131cea2c57f.md>)

[Implementation](<https://github.com/matejgrcic/DenseFlow>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### VDM

NLL (bits/dim): 3.4. Result ID: 698f13533a6c314849f5840d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f071e209131cea4365c.md>)

[Implementation](<https://github.com/google-research/vdm>)

Recorded evaluation configuration: {"data augmentation":true}. This record does not assert independent reproduction.

### Sparse Transformer

NLL (bits/dim): 3.44. Result ID: 698f13533a6c314849f5840c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65aa1e209131cea0bc26.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### SPN

NLL (bits/dim): 3.52. Result ID: 698f13533a6c314849f5840b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa651c1e209131cea086a4.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### VDVAE

NLL (bits/dim): 3.52. Result ID: 698f13533a6c314849f5840e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69931e209131cea22c7a.md>)

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

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Flow++

NLL (bits/dim): 3.69. Result ID: 698f13533a6c314849f58408.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

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

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

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### GLOW

NLL (bits/dim): 3.81. Result ID: 698f13533a6c314849f58409.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c6a1e209131cea33fb1.md>)

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

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

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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