Hugging Face's logo

qualcomm
/
OpenAI-Clip

Image Classification
PyTorch
foundation
android
Model card Files
xet
 Community
1
qaihm-bot commited on
Commit
6c14af9
·
verified ·
1 Parent(s): e23c4ef

v0.48.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.48.0 for changelog.

Files changed (1) hide show
  1. README.md +32 -32
README.md CHANGED
@@ -15,7 +15,7 @@ pipeline_tag: image-classification
15
  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.
16
 
17
  This is based on the implementation of OpenAI-Clip found [here](https://github.com/openai/CLIP/).
18
- This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
21
 
@@ -29,14 +29,14 @@ Download pre-exported model assets from **[OpenAI-Clip on Qualcomm® AI Hub](htt
29
 
30
  ### Option 2: Export with Custom Configurations
31
 
32
- Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) Python library to compile and export the model with your own:
33
  - Custom weights (e.g., fine-tuned checkpoints)
34
  - Custom input shapes
35
  - Target device and runtime configurations
36
 
37
  This option is ideal if you need to customize the model beyond the default configuration provided here.
38
 
39
- See our repository for [OpenAI-Clip on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) for usage instructions.
40
 
41
  ## Model Details
42
 
@@ -52,35 +52,35 @@ See our repository for [OpenAI-Clip on GitHub](https://github.com/quic/ai-hub-mo
52
  ## Performance Summary
53
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
54
  |---|---|---|---|---|---|---
55
- | OpenAI-Clip | ONNX | float | Snapdragon® X Elite | 16.457 ms | 291 - 291 MB | NPU
56
- | OpenAI-Clip | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 11.291 ms | 0 - 563 MB | NPU
57
- | OpenAI-Clip | ONNX | float | Qualcomm® QCS8550 (Proxy) | 15.753 ms | 0 - 322 MB | NPU
58
- | OpenAI-Clip | ONNX | float | Qualcomm® QCS9075 | 20.495 ms | 0 - 4 MB | NPU
59
- | OpenAI-Clip | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.064 ms | 1 - 534 MB | NPU
60
- | OpenAI-Clip | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.995 ms | 1 - 496 MB | NPU
61
- | OpenAI-Clip | ONNX | float | Snapdragon® X2 Elite | 7.22 ms | 291 - 291 MB | NPU
62
- | OpenAI-Clip | QNN_DLC | float | Snapdragon® X Elite | 18.869 ms | 1 - 1 MB | NPU
63
- | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 12.582 ms | 0 - 550 MB | NPU
64
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 56.011 ms | 1 - 506 MB | NPU
65
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 17.87 ms | 1 - 3 MB | NPU
66
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA8775P | 97.264 ms | 1 - 504 MB | NPU
67
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS9075 | 20.917 ms | 1 - 3 MB | NPU
68
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 20.991 ms | 0 - 503 MB | NPU
69
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA7255P | 56.011 ms | 1 - 506 MB | NPU
70
- | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA8295P | 22.079 ms | 1 - 496 MB | NPU
71
- | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.557 ms | 1 - 516 MB | NPU
72
- | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.333 ms | 0 - 486 MB | NPU
73
- | OpenAI-Clip | QNN_DLC | float | Snapdragon® X2 Elite | 9.078 ms | 1 - 1 MB | NPU
74
- | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 11.083 ms | 0 - 554 MB | NPU
75
- | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 52.04 ms | 0 - 509 MB | NPU
76
- | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 15.622 ms | 0 - 3 MB | NPU
77
- | OpenAI-Clip | TFLITE | float | Qualcomm® SA8775P | 18.658 ms | 0 - 507 MB | NPU
78
- | OpenAI-Clip | TFLITE | float | Qualcomm® QCS9075 | 20.687 ms | 0 - 294 MB | NPU
79
- | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 20.313 ms | 0 - 499 MB | NPU
80
- | OpenAI-Clip | TFLITE | float | Qualcomm® SA7255P | 52.04 ms | 0 - 509 MB | NPU
81
- | OpenAI-Clip | TFLITE | float | Qualcomm® SA8295P | 21.243 ms | 0 - 495 MB | NPU
82
- | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.007 ms | 0 - 514 MB | NPU
83
- | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.926 ms | 0 - 494 MB | NPU
84
 
85
  ## License
86
  * The license for the original implementation of OpenAI-Clip can be found
 
15
  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.
16
 
17
  This is based on the implementation of OpenAI-Clip found [here](https://github.com/openai/CLIP/).
18
+ This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
21
 
 
29
 
30
  ### Option 2: Export with Custom Configurations
31
 
32
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) Python library to compile and export the model with your own:
33
  - Custom weights (e.g., fine-tuned checkpoints)
34
  - Custom input shapes
35
  - Target device and runtime configurations
36
 
37
  This option is ideal if you need to customize the model beyond the default configuration provided here.
38
 
39
+ See our repository for [OpenAI-Clip on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/openai_clip) for usage instructions.
40
 
41
  ## Model Details
42
 
 
52
  ## Performance Summary
53
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
54
  |---|---|---|---|---|---|---
55
+ | OpenAI-Clip | ONNX | float | Snapdragon® X2 Elite | 7.205 ms | 291 - 291 MB | NPU
56
+ | OpenAI-Clip | ONNX | float | Snapdragon® X Elite | 16.481 ms | 291 - 291 MB | NPU
57
+ | OpenAI-Clip | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 11.304 ms | 1 - 564 MB | NPU
58
+ | OpenAI-Clip | ONNX | float | Qualcomm® QCS8550 (Proxy) | 15.726 ms | 0 - 335 MB | NPU
59
+ | OpenAI-Clip | ONNX | float | Qualcomm® QCS9075 | 20.388 ms | 0 - 4 MB | NPU
60
+ | OpenAI-Clip | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.086 ms | 1 - 533 MB | NPU
61
+ | OpenAI-Clip | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.999 ms | 1 - 496 MB | NPU
62
+ | OpenAI-Clip | QNN_DLC | float | Snapdragon® X2 Elite | 9.095 ms | 1 - 1 MB | NPU
63
+ | OpenAI-Clip | QNN_DLC | float | Snapdragon® X Elite | 18.846 ms | 1 - 1 MB | NPU
64
+ | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 12.55 ms | 0 - 554 MB | NPU
65
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 56.004 ms | 1 - 505 MB | NPU
66
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 17.857 ms | 1 - 3 MB | NPU
67
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA8775P | 20.945 ms | 1 - 504 MB | NPU
68
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS9075 | 21.217 ms | 3 - 5 MB | NPU
69
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 21.0 ms | 0 - 502 MB | NPU
70
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA7255P | 56.004 ms | 1 - 505 MB | NPU
71
+ | OpenAI-Clip | QNN_DLC | float | Qualcomm® SA8295P | 22.083 ms | 1 - 497 MB | NPU
72
+ | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.522 ms | 1 - 516 MB | NPU
73
+ | OpenAI-Clip | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.338 ms | 0 - 486 MB | NPU
74
+ | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 11.082 ms | 0 - 559 MB | NPU
75
+ | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 52.065 ms | 0 - 508 MB | NPU
76
+ | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 15.604 ms | 0 - 3 MB | NPU
77
+ | OpenAI-Clip | TFLITE | float | Qualcomm® SA8775P | 18.667 ms | 0 - 508 MB | NPU
78
+ | OpenAI-Clip | TFLITE | float | Qualcomm® QCS9075 | 20.359 ms | 0 - 294 MB | NPU
79
+ | OpenAI-Clip | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 20.309 ms | 0 - 502 MB | NPU
80
+ | OpenAI-Clip | TFLITE | float | Qualcomm® SA7255P | 52.065 ms | 0 - 508 MB | NPU
81
+ | OpenAI-Clip | TFLITE | float | Qualcomm® SA8295P | 21.252 ms | 0 - 495 MB | NPU
82
+ | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.021 ms | 0 - 517 MB | NPU
83
+ | OpenAI-Clip | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.928 ms | 0 - 496 MB | NPU
84
 
85
  ## License
86
  * The license for the original implementation of OpenAI-Clip can be found