scenario
stringlengths 33
60
| model_cls
stringclasses 4
values | num_params_B
float64 2.57
14.3
| flops_G
float64 5.98k
786k
⌀ | time_plain_s
stringlengths 12
15
| mem_plain_GB
float64 0.2
30.3
| time_compile_s
stringclasses 4
values | mem_compile_GB
float64 5.24
31.8
⌀ | fullgraph
bool 1
class | mode
stringclasses 1
value | github_sha
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|
black-forest-labs/FLUX.1-dev-bf16
|
FluxTransformer2DModel
| 11.9
| 59,523.04
|
0.535 (0.537)
| 22.63
|
0.391 (0.395)
| 22.73
| true
|
default
|
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
black-forest-labs/FLUX.1-dev-bnb-nf4
|
FluxTransformer2DModel
| 5.95
| null |
0.564 (0.57)
| 6.72
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
black-forest-labs/FLUX.1-dev-layerwise-upcasting
|
FluxTransformer2DModel
| 11.9
| 59,523.04
|
0.599 (0.602)
| 22.18
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
black-forest-labs/FLUX.1-dev-group-offload-leaf
|
FluxTransformer2DModel
| 11.9
| 59,523.04
|
1.879 (1.89)
| 0.55
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Lightricks/LTX-Video-0.9.7-dev-bf16
|
LTXVideoTransformer3DModel
| 13.04
| 167,565.8
|
1.656 (1.654)
| 25.22
|
1.143 (1.135)
| 25.31
| true
|
default
|
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Lightricks/LTX-Video-0.9.7-dev-layerwise-upcasting
|
LTXVideoTransformer3DModel
| 13.04
| 167,565.8
|
1.714 (1.716)
| 24.38
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Lightricks/LTX-Video-0.9.7-dev-group-offload-leaf
|
LTXVideoTransformer3DModel
| 13.04
| 167,565.8
|
2.795 (2.81)
| 1.05
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
stabilityai/stable-diffusion-xl-base-1.0-bf16
|
UNet2DConditionModel
| 2.57
| 5,979.1
|
0.067 (0.073)
| 5.05
|
0.054
| 5.24
| true
|
default
|
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
stabilityai/stable-diffusion-xl-base-1.0-layerwise-upcasting
|
UNet2DConditionModel
| 2.57
| 5,979.1
|
0.127 (0.146)
| 4.89
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
stabilityai/stable-diffusion-xl-base-1.0-group-offload-leaf
|
UNet2DConditionModel
| 2.57
| 5,979.1
|
0.49 (0.506)
| 0.2
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Wan-AI/Wan2.1-T2V-14B-Diffusers-bf16
|
WanTransformer3DModel
| 14.29
| 785,570.99
|
11.517 (11.443)
| 30.25
|
8.793 (8.895)
| 31.78
| true
|
default
|
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Wan-AI/Wan2.1-T2V-14B-Diffusers-layerwise-upcasting
|
WanTransformer3DModel
| 14.29
| 785,570.99
|
11.305 (11.592)
| 26.78
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Wan-AI/Wan2.1-T2V-14B-Diffusers-group-offload-leaf
|
WanTransformer3DModel
| 14.29
| 785,570.99
|
12.02 (11.745)
| 3.54
| null | null | null | null |
b4e6dc3037e75e2dc16466b914b1be597b06be9a
|
Welcome to 🤗 Diffusers Benchmarks!
This is dataset where we keep track of the inference latency and memory information of the core models in the diffusers library.
Currently, the core models are:
- Flux
- Wan
- LTX
- SDXL
Note that we will continue to extend this list based on their usage.
You can analyze the results in this demo.
Instead of benchmarking the entire diffusion pipelines, we only benchmark the forward passes of the diffusion networks under different settings (compile, offloading, quantization, etc.). We only benchmark the diffusion networks because they are typically the most compute-heavy part in a diffusion pipeline.
We use this GitHub Actions workflow to report the above numbers automatically. This workflow runs on a biweekly cadence.
Here is an example run.
- Downloads last month
- 47