pharmapsychotic commited on
Commit
ed45558
1 Parent(s): 942ef5e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -8
README.md CHANGED
@@ -12,14 +12,14 @@ tags:
12
 
13
  # Stable Diffusion XL 1.0 TensorRT
14
 
15
- ### Introduction
16
 
17
  This repository hosts the TensorRT versions of **Stable Diffusion XL 1.0** created in collaboration with [NVIDIA](https://huggingface.co/nvidia). The optimized versions give substantial improvements in speed and efficiency.
18
 
19
 
20
  ![examples](./examples.jpg)
21
 
22
- ### Model Description
23
 
24
  - **Developed by:** Stability AI
25
  - **Model type:** Diffusion-based text-to-image generative model
@@ -27,7 +27,7 @@ This repository hosts the TensorRT versions of **Stable Diffusion XL 1.0** creat
27
  - **Model Description:** This is a conversion of the [SDXL base 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) and [SDXL refiner 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0) models for [NVIDIA TensorRT](https://developer.nvidia.com/tensorrt) optimized inference
28
 
29
 
30
- ### Performance Comparison
31
 
32
  #### Timings for 30 steps at 1024x1024
33
 
@@ -37,7 +37,7 @@ This repository hosts the TensorRT versions of **Stable Diffusion XL 1.0** creat
37
  | A100 | 3704 ms | 2742 ms | ~26% |
38
  | H100 | 2496 ms | 1471 ms | ~41% |
39
 
40
- #### Image throughput for 30 steps
41
 
42
  | Accelerator | Baseline (non-optimized) | NVIDIA TensorRT (optimized) | Percentage improvement |
43
  |-------------|--------------------------|-----------------------------|------------------------|
@@ -46,7 +46,7 @@ This repository hosts the TensorRT versions of **Stable Diffusion XL 1.0** creat
46
  | H100 | 0.40 images/sec | 0.68 images/sec | ~70% |
47
 
48
 
49
- ### Usage Example
50
 
51
  1. Following the [setup instructions](https://github.com/rajeevsrao/TensorRT/blob/release/8.6/demo/Diffusion/README.md) for TensorRT on launching a TensorRT NGC container.
52
  ```shell
@@ -70,8 +70,7 @@ cd ..
70
  python3 -m pip install --upgrade pip
71
  python3 -m pip install --upgrade tensorrt
72
 
73
- export TRT_OSSPATH=/workspace
74
- cd $TRT_OSSPATH/demo/Diffusion
75
  pip3 install -r requirements.txt
76
  ```
77
 
@@ -79,7 +78,6 @@ pip3 install -r requirements.txt
79
  ```
80
  python3 demo_txt2img_xl.py \
81
  "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" \
82
- --hf-token=<Your HF TOKEN> \
83
  --build-static-batch \
84
  --use-cuda-graph \
85
  --num-warmup-runs 1 \
 
12
 
13
  # Stable Diffusion XL 1.0 TensorRT
14
 
15
+ ## Introduction
16
 
17
  This repository hosts the TensorRT versions of **Stable Diffusion XL 1.0** created in collaboration with [NVIDIA](https://huggingface.co/nvidia). The optimized versions give substantial improvements in speed and efficiency.
18
 
19
 
20
  ![examples](./examples.jpg)
21
 
22
+ ## Model Description
23
 
24
  - **Developed by:** Stability AI
25
  - **Model type:** Diffusion-based text-to-image generative model
 
27
  - **Model Description:** This is a conversion of the [SDXL base 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) and [SDXL refiner 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0) models for [NVIDIA TensorRT](https://developer.nvidia.com/tensorrt) optimized inference
28
 
29
 
30
+ ## Performance Comparison
31
 
32
  #### Timings for 30 steps at 1024x1024
33
 
 
37
  | A100 | 3704 ms | 2742 ms | ~26% |
38
  | H100 | 2496 ms | 1471 ms | ~41% |
39
 
40
+ #### Image throughput for 30 steps at 1024x1024
41
 
42
  | Accelerator | Baseline (non-optimized) | NVIDIA TensorRT (optimized) | Percentage improvement |
43
  |-------------|--------------------------|-----------------------------|------------------------|
 
46
  | H100 | 0.40 images/sec | 0.68 images/sec | ~70% |
47
 
48
 
49
+ ## Usage Example
50
 
51
  1. Following the [setup instructions](https://github.com/rajeevsrao/TensorRT/blob/release/8.6/demo/Diffusion/README.md) for TensorRT on launching a TensorRT NGC container.
52
  ```shell
 
70
  python3 -m pip install --upgrade pip
71
  python3 -m pip install --upgrade tensorrt
72
 
73
+ cd demo/Diffusion
 
74
  pip3 install -r requirements.txt
75
  ```
76
 
 
78
  ```
79
  python3 demo_txt2img_xl.py \
80
  "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" \
 
81
  --build-static-batch \
82
  --use-cuda-graph \
83
  --num-warmup-runs 1 \