amd
/

Text-to-Image
Diffusers
akasharidas commited on
Commit
1a94ed8
·
verified ·
1 Parent(s): bbab59d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -7
README.md CHANGED
@@ -7,16 +7,16 @@ base_model:
7
  pipeline_tag: text-to-image
8
  library_name: diffusers
9
  ---
10
- # AMD Nitro Diffusion
11
 
12
 
13
  ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6355aded9c72a7e742f341a4/AsUvS7acUDLZhKOMRSH37.jpeg)
14
 
15
  ## Introduction
16
- AMD Nitro Diffusion is a series of efficient text-to-image generation models that are distilled from popular diffusion models on AMD Instinct™ GPUs. The release consists of:
17
 
18
- * [Stable Diffusion 2.1 Nitro](https://huggingface.co/amd/SD2.1-Nitro): a UNet-based one-step model distilled from [Stable Diffusion 2.1](https://huggingface.co/stabilityai/stable-diffusion-2-1-base).
19
- * [PixArt-Sigma Nitro](https://huggingface.co/amd/PixArt-Sigma-Nitro): a high resolution transformer-based one-step model distilled from [PixArt-Sigma](https://pixart-alpha.github.io/PixArt-sigma-project/).
20
 
21
  ⚡️ [Open-source code](https://github.com/AMD-AIG-AIMA/AMD-Diffusion-Distillation)! The models are based on our re-implementation of [Latent Adversarial Diffusion Distillation](https://arxiv.org/abs/2403.12015), the method used to build the popular Stable Diffusion 3 Turbo model. Since the original authors didn't provide training code, we release our re-implementation to help advance further research in the field.
22
 
@@ -24,9 +24,9 @@ AMD Nitro Diffusion is a series of efficient text-to-image generation models tha
24
 
25
  ## Details
26
 
27
- * **Model architecture**: PixArt-Sigma Nitro has the same architecture as PixArt-Sigma and is compatible with the diffusers pipeline.
28
  * **Inference steps**: This model is distilled to perform inference in just a single step. However, the training code also supports distilling a model for 2, 4 or 8 steps.
29
- * **Hardware**: We use a single node consisting of 4 AMD Instinct™ MI250 GPUs for distilling PixArt-Sigma Nitro.
30
  * **Dataset**: We use 1M prompts from [DiffusionDB](https://huggingface.co/datasets/poloclub/diffusiondb) and generate the corresponding images from the base PixArt-Sigma model.
31
  * **Training cost**: The distillation process achieves reasonable results in less than 2 days on a single node.
32
 
@@ -64,7 +64,7 @@ Compared to [PixArt-Sigma](https://pixart-alpha.github.io/PixArt-sigma-project/)
64
  | Model | FID ↓ | CLIP ↑ |FLOPs| Latency on AMD Instinct MI250 (sec)
65
  | :---: | :---: | :---: | :---: | :---:
66
  | PixArt-Sigma, 20 steps | 34.14 | 0.3289 |187.96 | 7.46
67
- | **PixArt-Sigma Nitro**, 1 step | 37.75 | 0.3167|17.04 | 0.53
68
 
69
 
70
 
 
7
  pipeline_tag: text-to-image
8
  library_name: diffusers
9
  ---
10
+ # AMD Nitro-1
11
 
12
 
13
  ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6355aded9c72a7e742f341a4/AsUvS7acUDLZhKOMRSH37.jpeg)
14
 
15
  ## Introduction
16
+ Nitro-1 is a series of efficient text-to-image generation models that are distilled from popular diffusion models on AMD Instinct™ GPUs. The release consists of:
17
 
18
+ * [Nitro-1-SD](https://huggingface.co/amd/SD2.1-Nitro): a UNet-based one-step model distilled from [Stable Diffusion 2.1](https://huggingface.co/stabilityai/stable-diffusion-2-1-base).
19
+ * [Nitro-1-PixArt](https://huggingface.co/amd/PixArt-Sigma-Nitro): a high resolution transformer-based one-step model distilled from [PixArt-Sigma](https://pixart-alpha.github.io/PixArt-sigma-project/).
20
 
21
  ⚡️ [Open-source code](https://github.com/AMD-AIG-AIMA/AMD-Diffusion-Distillation)! The models are based on our re-implementation of [Latent Adversarial Diffusion Distillation](https://arxiv.org/abs/2403.12015), the method used to build the popular Stable Diffusion 3 Turbo model. Since the original authors didn't provide training code, we release our re-implementation to help advance further research in the field.
22
 
 
24
 
25
  ## Details
26
 
27
+ * **Model architecture**: Nitro-1-PixArt has the same architecture as PixArt-Sigma and is compatible with the diffusers pipeline.
28
  * **Inference steps**: This model is distilled to perform inference in just a single step. However, the training code also supports distilling a model for 2, 4 or 8 steps.
29
+ * **Hardware**: We use a single node consisting of 4 AMD Instinct™ MI250 GPUs for distilling Nitro-1-PixArt.
30
  * **Dataset**: We use 1M prompts from [DiffusionDB](https://huggingface.co/datasets/poloclub/diffusiondb) and generate the corresponding images from the base PixArt-Sigma model.
31
  * **Training cost**: The distillation process achieves reasonable results in less than 2 days on a single node.
32
 
 
64
  | Model | FID ↓ | CLIP ↑ |FLOPs| Latency on AMD Instinct MI250 (sec)
65
  | :---: | :---: | :---: | :---: | :---:
66
  | PixArt-Sigma, 20 steps | 34.14 | 0.3289 |187.96 | 7.46
67
+ | **Nitro-1-PixArt**, 1 step | 37.75 | 0.3167|17.04 | 0.53
68
 
69
 
70