Important
You are viewing the NeMo 2.0 documentation. This release introduces significant changes to the API and a new library, NeMo Run. We are currently porting all features from NeMo 1.0 to 2.0. For documentation on previous versions or features not yet available in 2.0, please refer to the NeMo 24.07 documentation.
Performance#
Training Quality Results#
Here we show some interesting results as an example of the dreambooth script.
Prompt: A ‘sks’ dog in a bucket.

Prompt: A ‘sks’ dog in Acropolis.

Prompt: A ‘sks’ dog in front of Eiffel tower.

Prompt: A ‘sks’ dog mecha robot.

The original source of images used for the above results are from link and is subject to the following license.
Inference Performance Results#
Latency times are started directly before the text encoding (CLIP) and stopped directly after the output image decoding (VAE). For framework we use the Torch Automated Mixed Precision (AMP) for FP16 computation. For TRT, we export the various models with the FP16 acceleration. We use the optimized TRT engine setup present in the deployment directory to get the numbers in the same environment as the framework.
GPU: NVIDIA DGX A100 (1x A100 80 GB)
Batch Size: Synonymous with num_images_per_prompt
Model |
Batch Size |
Sampler |
Inference Steps |
TRT FP 16 Latency (s) |
FW FP 16 (AMP) Latency (s) |
TRT vs FW Speedup (x) |
---|---|---|---|---|---|---|
DreamBooth (Res=256) |
1 |
DDIM |
100 |
2.0 |
5.6 |
2.8 |
DreamBooth (Res=256) |
2 |
DDIM |
100 |
3.1 |
9.0 |
2.9 |
DreamBooth (Res=256) |
4 |
DDIM |
100 |
5.7 |
16.0 |
2.8 |