Hugging Face's logo

qualcomm
/
OpenAI-Clip

Image Classification
PyTorch
foundation
android
Model card Files
xet
 Community
1
OpenAI-Clip / README.md
qaihm-bot's picture
qaihm-bot
v0.30.5
45c4e21 verified
preview code
|
raw
history blame
12.3 kB
metadata 
library_name: pytorch
license: other
tags:
  - foundation
  - android
pipeline_tag: image-classification

OpenAI-Clip: Optimized for Mobile Deployment

Multi-modal foundational model for vision and language tasks like image/text similarity and for zero-shot image classification

Contrastive Language-Image Pre-Training (CLIP) uses a ViT like transformer to get visual features and a causal language model to get the text features. Both the text and visual features can then be used for a variety of zero-shot learning tasks.

This model is an implementation of OpenAI-Clip found here.

This repository provides scripts to run OpenAI-Clip on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

  • Model Type: Model_use_case.image_classification
  • Model Stats:
    • Model checkpoint: ViT-B/16
    • Image input resolution: 224x224
    • Text context length: 77
    • Number of parameters (CLIPTextEncoder): 76.0M
    • Model size (CLIPTextEncoder): 290 MB
    • Number of parameters (CLIPImageEncoder): 115M
    • Model size (CLIPImageEncoder): 437 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
OpenAI-Clip float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 367.248 ms 0 - 406 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 305.317 ms 1 - 10 MB NPU Use Export Script
OpenAI-Clip float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 26.29 ms 0 - 364 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 22.919 ms 1 - 455 MB NPU Use Export Script
OpenAI-Clip float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 23.813 ms 0 - 48 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 19.969 ms 1 - 3 MB NPU Use Export Script
OpenAI-Clip float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 28.479 ms 0 - 407 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 22.717 ms 1 - 10 MB NPU Use Export Script
OpenAI-Clip float SA7255P ADP Qualcomm® SA7255P TFLITE 367.248 ms 0 - 406 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float SA7255P ADP Qualcomm® SA7255P QNN 305.317 ms 1 - 10 MB NPU Use Export Script
OpenAI-Clip float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 23.865 ms 0 - 43 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 19.894 ms 1 - 3 MB NPU Use Export Script
OpenAI-Clip float SA8295P ADP Qualcomm® SA8295P TFLITE 29.633 ms 0 - 349 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float SA8295P ADP Qualcomm® SA8295P QNN 23.848 ms 1 - 15 MB NPU Use Export Script
OpenAI-Clip float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 23.649 ms 0 - 63 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 19.842 ms 1 - 8 MB NPU Use Export Script
OpenAI-Clip float SA8775P ADP Qualcomm® SA8775P TFLITE 28.479 ms 0 - 407 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float SA8775P ADP Qualcomm® SA8775P QNN 22.717 ms 1 - 10 MB NPU Use Export Script
OpenAI-Clip float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 23.99 ms 0 - 53 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 19.815 ms 0 - 47 MB NPU Use Export Script
OpenAI-Clip float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 24.691 ms 0 - 54 MB NPU OpenAI-Clip.onnx
OpenAI-Clip float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 16.648 ms 0 - 414 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 14.178 ms 1 - 491 MB NPU Use Export Script
OpenAI-Clip float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 17.654 ms 0 - 532 MB NPU OpenAI-Clip.onnx
OpenAI-Clip float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 15.935 ms 0 - 406 MB NPU OpenAI-Clip.tflite
OpenAI-Clip float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 13.223 ms 1 - 470 MB NPU Use Export Script
OpenAI-Clip float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 16.872 ms 1 - 512 MB NPU OpenAI-Clip.onnx
OpenAI-Clip float Snapdragon X Elite CRD Snapdragon® X Elite QNN 20.798 ms 1 - 1 MB NPU Use Export Script
OpenAI-Clip float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 25.8 ms 293 - 293 MB NPU OpenAI-Clip.onnx

Installation

Install the package via pip:

pip install "qai-hub-models[openai-clip]"

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

Sign-in to Qualcomm® AI Hub 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.

qai-hub configure --api_token API_TOKEN

Navigate to 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.

python -m qai_hub_models.models.openai_clip.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.openai_clip.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.
python -m qai_hub_models.models.openai_clip.export

Profiling Results
------------------------------------------------------------
OpenAI-Clip
Device                          : cs_8275 (ANDROID 14)                  
Runtime                         : TFLITE                                
Estimated inference time (ms)   : 367.2                                 
Estimated peak memory usage (MB): [0, 406]                              
Total # Ops                     : 1320                                  
Compute Unit(s)                 : npu (1318 ops) gpu (0 ops) cpu (2 ops)

How does this work?

This export script leverages Qualcomm® AI Hub to optimize, validate, and deploy this model on-device. Lets go through each step below in detail:

Step 1: Compile model for on-device deployment

To compile a PyTorch model for on-device deployment, we first trace the model in memory using the jit.trace and then call the submit_compile_job API.

import torch

import qai_hub as hub
from qai_hub_models.models.openai_clip import Model

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S24")

# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()

pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])

# Compile model on a specific device
compile_job = hub.submit_compile_job(
    model=pt_model,
    device=device,
    input_specs=torch_model.get_input_spec(),
)

# Get target model to run on-device
target_model = compile_job.get_target_model()

Step 2: Performance profiling on cloud-hosted device

After compiling models from step 1. Models can be profiled model on-device using the target_model. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics.

profile_job = hub.submit_profile_job(
    model=target_model,
    device=device,
)

Step 3: Verify on-device accuracy

To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device.

input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = inference_job.download_output_data()

With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output.

Note: This on-device profiling and inference requires access to Qualcomm® AI Hub. Sign up for access.

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

  • TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.

  • QNN (.so export ): This sample app provides instructions on how to use the .so shared library in an Android application.

View on Qualcomm® AI Hub

Get more details on OpenAI-Clip's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

  • The license for the original implementation of OpenAI-Clip can be found here.
  • The license for the compiled assets for on-device deployment can be found here

References

Community