Upload README.md with huggingface_hub

8967fdc verified 9 months ago

13.6 kB

	---
	library_name: pytorch
	license: other
	tags:
	- generative_ai
	- android
	pipeline_tag: unconditional-image-generation

	---

	![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/stable_diffusion_v2_1_quantized/web-assets/model_demo.png)

	# Stable-Diffusion-v2.1: Optimized for Mobile Deployment
	## State-of-the-art generative AI model used to generate detailed images conditioned on text descriptions


	Generates high resolution images from text prompts using a latent diffusion model. This model uses CLIP ViT-L/14 as text encoder, U-Net based latent denoising, and VAE based decoder to generate the final image.

	This model is an implementation of Stable-Diffusion-v2.1 found [here](https://github.com/CompVis/stable-diffusion/tree/main).


	This repository provides scripts to run Stable-Diffusion-v2.1 on Qualcomm® devices.
	More details on model performance across various devices, can be found
	[here](https://aihub.qualcomm.com/models/stable_diffusion_v2_1_quantized).


	### Model Details

	- Model Type: Model_use_case.image_generation
	- Model Stats:
	- Input: Text prompt to generate image
	- Text Encoder Number of parameters: 340M
	- UNet Number of parameters: 865M
	- VAE Decoder Number of parameters: 83M
	- Model size: 1GB

	\| Model \| Precision \| Device \| Chipset \| Target Runtime \| Inference Time (ms) \| Peak Memory Range (MB) \| Primary Compute Unit \| Target Model
	\|---\|---\|---\|---\|---\|---\|---\|---\|---\|
	\| TextEncoderQuantizable \| w8a16 \| QCS8275 (Proxy) \| Qualcomm® QCS8275 (Proxy) \| QNN \| 15.92 ms \| 0 - 9 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| QCS8550 (Proxy) \| Qualcomm® QCS8550 (Proxy) \| QNN \| 6.594 ms \| 0 - 2 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| QCS9075 (Proxy) \| Qualcomm® QCS9075 (Proxy) \| QNN \| 6.814 ms \| 0 - 9 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| SA7255P ADP \| Qualcomm® SA7255P \| QNN \| 15.92 ms \| 0 - 9 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| SA8255 (Proxy) \| Qualcomm® SA8255P (Proxy) \| QNN \| 6.634 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| SA8650 (Proxy) \| Qualcomm® SA8650P (Proxy) \| QNN \| 6.813 ms \| 0 - 2 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| SA8775P ADP \| Qualcomm® SA8775P \| QNN \| 6.814 ms \| 0 - 9 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| QNN \| 6.632 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| ONNX \| 6.859 ms \| 0 - 386 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| TextEncoderQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| QNN \| 4.62 ms \| 0 - 19 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| ONNX \| 5.075 ms \| 0 - 14 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| TextEncoderQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| QNN \| 4.173 ms \| 0 - 14 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| ONNX \| 4.499 ms \| 0 - 13 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| TextEncoderQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| QNN \| 6.834 ms \| 0 - 0 MB \| NPU \| Use Export Script \|
	\| TextEncoderQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| ONNX \| 6.823 ms \| 378 - 378 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| UnetQuantizable \| w8a16 \| QCS8275 (Proxy) \| Qualcomm® QCS8275 (Proxy) \| QNN \| 241.358 ms \| 0 - 8 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| QCS8550 (Proxy) \| Qualcomm® QCS8550 (Proxy) \| QNN \| 97.192 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| QCS9075 (Proxy) \| Qualcomm® QCS9075 (Proxy) \| QNN \| 92.111 ms \| 0 - 8 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| SA7255P ADP \| Qualcomm® SA7255P \| QNN \| 241.358 ms \| 0 - 8 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| SA8255 (Proxy) \| Qualcomm® SA8255P (Proxy) \| QNN \| 97.903 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| SA8650 (Proxy) \| Qualcomm® SA8650P (Proxy) \| QNN \| 96.775 ms \| 0 - 2 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| SA8775P ADP \| Qualcomm® SA8775P \| QNN \| 92.111 ms \| 0 - 8 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| QNN \| 97.437 ms \| 0 - 5 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| ONNX \| 98.551 ms \| 0 - 3 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| UnetQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| QNN \| 69.198 ms \| 0 - 19 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| ONNX \| 69.745 ms \| 0 - 17 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| UnetQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| QNN \| 61.688 ms \| 0 - 14 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| ONNX \| 62.663 ms \| 0 - 14 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| UnetQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| QNN \| 98.945 ms \| 0 - 0 MB \| NPU \| Use Export Script \|
	\| UnetQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| ONNX \| 99.538 ms \| 842 - 842 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| VaeDecoderQuantizable \| w8a16 \| QCS8275 (Proxy) \| Qualcomm® QCS8275 (Proxy) \| QNN \| 720.835 ms \| 1 - 10 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| QCS8550 (Proxy) \| Qualcomm® QCS8550 (Proxy) \| QNN \| 270.663 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| QCS9075 (Proxy) \| Qualcomm® QCS9075 (Proxy) \| QNN \| 250.403 ms \| 0 - 12 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| SA7255P ADP \| Qualcomm® SA7255P \| QNN \| 720.835 ms \| 1 - 10 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| SA8255 (Proxy) \| Qualcomm® SA8255P (Proxy) \| QNN \| 271.924 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| SA8650 (Proxy) \| Qualcomm® SA8650P (Proxy) \| QNN \| 274.917 ms \| 0 - 2 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| SA8775P ADP \| Qualcomm® SA8775P \| QNN \| 250.403 ms \| 0 - 12 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| QNN \| 269.888 ms \| 0 - 3 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| Samsung Galaxy S23 \| Snapdragon® 8 Gen 2 Mobile \| ONNX \| 269.758 ms \| 0 - 66 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| VaeDecoderQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| QNN \| 203.298 ms \| 0 - 20 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| Samsung Galaxy S24 \| Snapdragon® 8 Gen 3 Mobile \| ONNX \| 204.412 ms \| 3 - 22 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| VaeDecoderQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| QNN \| 193.019 ms \| 0 - 15 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| Snapdragon 8 Elite QRD \| Snapdragon® 8 Elite Mobile \| ONNX \| 188.783 ms \| 3 - 17 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|
	\| VaeDecoderQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| QNN \| 266.283 ms \| 0 - 0 MB \| NPU \| Use Export Script \|
	\| VaeDecoderQuantizable \| w8a16 \| Snapdragon X Elite CRD \| Snapdragon® X Elite \| ONNX \| 266.925 ms \| 63 - 63 MB \| NPU \| [Stable-Diffusion-v2.1.onnx](https://huggingface.co/qualcomm/Stable-Diffusion-v2.1/blob/main/Stable-Diffusion-v2.1_w8a16.onnx) \|




	## Installation


	Install the package via pip:
	```bash
	pip install "qai-hub-models[stable-diffusion-v2-1-quantized]"
	```


	## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

	Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
	Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.

	With this API token, you can configure your client to run models on the cloud
	hosted devices.
	```bash
	qai-hub configure --api_token API_TOKEN
	```
	Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.



	## Demo off target

	The package contains a simple end-to-end demo that downloads pre-trained
	weights and runs this model on a sample input.

	```bash
	python -m qai_hub_models.models.stable_diffusion_v2_1_quantized.demo
	```

	The above demo runs a reference implementation of pre-processing, model
	inference, and post processing.

	NOTE: If you want running in a Jupyter Notebook or Google Colab like
	environment, please add the following to your cell (instead of the above).
	```
	%run -m qai_hub_models.models.stable_diffusion_v2_1_quantized.demo
	```


	### Run model on a cloud-hosted device

	In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
	device. This script does the following:
	* Performance check on-device on a cloud-hosted device
	* Downloads compiled assets that can be deployed on-device for Android.
	* Accuracy check between PyTorch and on-device outputs.

	```bash
	python -m qai_hub_models.models.stable_diffusion_v2_1_quantized.export
	```
	```
	Profiling Results
	------------------------------------------------------------
	TextEncoderQuantizable
	Device : cs_8275 (ANDROID 14)
	Runtime : QNN
	Estimated inference time (ms) : 15.9
	Estimated peak memory usage (MB): [0, 9]
	Total # Ops : 971
	Compute Unit(s) : npu (971 ops) gpu (0 ops) cpu (0 ops)

	------------------------------------------------------------
	UnetQuantizable
	Device : cs_8275 (ANDROID 14)
	Runtime : QNN
	Estimated inference time (ms) : 241.4
	Estimated peak memory usage (MB): [0, 8]
	Total # Ops : 5783
	Compute Unit(s) : npu (5783 ops) gpu (0 ops) cpu (0 ops)

	------------------------------------------------------------
	VaeDecoderQuantizable
	Device : cs_8275 (ANDROID 14)
	Runtime : QNN
	Estimated inference time (ms) : 720.8
	Estimated peak memory usage (MB): [1, 10]
	Total # Ops : 189
	Compute Unit(s) : npu (189 ops) gpu (0 ops) cpu (0 ops)
	```





	## Deploying compiled model to Android


	The models can be deployed using multiple runtimes:
	- TensorFlow Lite (`.tflite` export): [This
	tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
	guide to deploy the .tflite model in an Android application.


	- QNN (`.so` export ): This [sample
	app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
	provides instructions on how to use the `.so` shared library in an Android application.


	## View on Qualcomm® AI Hub
	Get more details on Stable-Diffusion-v2.1's performance across various devices [here](https://aihub.qualcomm.com/models/stable_diffusion_v2_1_quantized).
	Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)


	## License
	* The license for the original implementation of Stable-Diffusion-v2.1 can be found
	[here](https://github.com/CompVis/stable-diffusion/blob/main/LICENSE).
	* The license for the compiled assets for on-device deployment can be found [here](https://github.com/CompVis/stable-diffusion/blob/main/LICENSE)



	## References
	* [High-Resolution Image Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752)
	* [Source Model Implementation](https://github.com/CompVis/stable-diffusion/tree/main)



	## Community
	* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
	* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).