Hugging Face's logo

qualcomm
/
MiniLM-v2

Text Generation
PyTorch
foundation
real_time
android
Model card Files
xet
 Community
MiniLM-v2
File size: 13,330 Bytes
7218b03
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
567d812
7218b03
 
 
 
 
 
 
 
 
 
 
 
567d812
 
 
 
 
 
7218b03
 
 
 
 
 
567d812
7218b03
 
 
 
 
 
567d812
7218b03
 
 
 
 
 
567d812
7218b03
567d812
7218b03
567d812
7218b03
 
 
 
567d812
 
 
 
 
 
 
 
 
 
 
 
f4752d0
567d812
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f4752d0
567d812
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7218b03
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
---
library_name: pytorch
license: other
tags:
- foundation
- real_time
- android
pipeline_tag: text-generation

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/web-assets/model_demo.png)

# MiniLM-v2: Optimized for Qualcomm Devices

All-MiniLM-L6-v2 maps sentences to a 384-dimensional dense vector space. Trained on 1B+ sentence pairs, it excels at semantic search, clustering, and sentence similarity tasks while being small enough to run on mobile devices.

This is based on the implementation of MiniLM-v2 found [here](https://github.com/UKPLab/sentence-transformers).
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/v0.60.0/src/qai_hub_models/models/minilm_v2) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

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.

## Getting Started
There are two ways to deploy this model on your device:

### Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.60.0/minilm_v2-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[MiniLM-v2 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/minilm_v2)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/minilm_v2) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for [MiniLM-v2 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/minilm_v2) for usage instructions.

## Model Details

**Model Type:** Model_use_case.text_generation

**Model Stats:**
- Embedding dimension: 384
- Input resolution: 128 tokens
- Model checkpoint: sentence-transformers/all-MiniLM-L6-v2
- Model size (float): 86.7 MB
- Number of parameters: 22.7M

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| MiniLM-v2 | ONNX | float | Snapdragon® X2 Elite | 0.672 ms | 1 - 1 MB | NPU
| MiniLM-v2 | ONNX | float | Snapdragon® X Elite | 1.676 ms | 44 - 44 MB | NPU
| MiniLM-v2 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.126 ms | 0 - 104 MB | NPU
| MiniLM-v2 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 2.357 ms | 0 - 107 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 1.784 ms | 0 - 4 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.615 ms | 0 - 50 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® QCS8450 | 2.357 ms | 0 - 107 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 2.437 ms | 0 - 3 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.676 ms | 44 - 44 MB | NPU
| MiniLM-v2 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.79 ms | 0 - 59 MB | NPU
| MiniLM-v2 | ONNX | float | Snapdragon® 8 Elite Mobile | 0.79 ms | 0 - 59 MB | NPU
| MiniLM-v2 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.643 ms | 0 - 61 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.672 ms | 1 - 1 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® X Elite | 1.783 ms | 23 - 23 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.194 ms | 0 - 84 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.98 ms | 0 - 87 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 5.535 ms | 0 - 3 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 1.67 ms | 0 - 4 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.715 ms | 0 - 26 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS8450 | 1.98 ms | 0 - 87 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 2.062 ms | 0 - 3 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.783 ms | 23 - 23 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 6.687 ms | 0 - 60 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.95 ms | 0 - 54 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.839 ms | 0 - 58 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.839 ms | 0 - 58 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.662 ms | 0 - 60 MB | NPU
| MiniLM-v2 | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.95 ms | 0 - 54 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® X2 Elite | 0.65 ms | 1 - 1 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® X Elite | 1.303 ms | 1 - 1 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.793 ms | 0 - 99 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.89 ms | 0 - 102 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 1.326 ms | 0 - 2 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 3.563 ms | 0 - 57 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.124 ms | 0 - 3 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® SA8775P | 1.51 ms | 0 - 59 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® SA8650P | 1.51 ms | 0 - 59 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® SA8255P | 1.51 ms | 0 - 59 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS8450 | 1.89 ms | 0 - 102 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 1.47 ms | 2 - 4 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.303 ms | 1 - 1 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.568 ms | 0 - 60 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® SA7255P | 3.563 ms | 0 - 57 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Qualcomm® SA8295P | 2.281 ms | 0 - 57 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 0.568 ms | 0 - 60 MB | NPU
| MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.466 ms | 0 - 61 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.806 ms | 1 - 1 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.909 ms | 0 - 0 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.205 ms | 0 - 74 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.943 ms | 0 - 72 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.065 ms | 0 - 2 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 1.657 ms | 0 - 2 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 3.424 ms | 0 - 48 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.726 ms | 0 - 33 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA8775P | 1.941 ms | 0 - 52 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA8650P | 1.941 ms | 0 - 52 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA8255P | 1.941 ms | 0 - 52 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 1.943 ms | 0 - 72 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.824 ms | 0 - 2 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.909 ms | 0 - 0 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 6.133 ms | 0 - 50 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.69 ms | 0 - 47 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.834 ms | 0 - 48 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 3.424 ms | 0 - 48 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 2.382 ms | 0 - 46 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.834 ms | 0 - 48 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.68 ms | 0 - 52 MB | NPU
| MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.69 ms | 0 - 47 MB | NPU
| MiniLM-v2 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.832 ms | 0 - 105 MB | NPU
| MiniLM-v2 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.952 ms | 0 - 101 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 1.401 ms | 0 - 46 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 3.732 ms | 0 - 63 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.127 ms | 0 - 2 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® SA8775P | 1.568 ms | 0 - 65 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® SA8650P | 1.568 ms | 0 - 65 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® SA8255P | 1.568 ms | 0 - 65 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® QCS8450 | 1.952 ms | 0 - 101 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.414 ms | 0 - 45 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.596 ms | 0 - 66 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® SA7255P | 3.732 ms | 0 - 63 MB | NPU
| MiniLM-v2 | TFLITE | float | Qualcomm® SA8295P | 2.362 ms | 0 - 56 MB | NPU
| MiniLM-v2 | TFLITE | float | Snapdragon® 8 Elite Mobile | 0.596 ms | 0 - 66 MB | NPU
| MiniLM-v2 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.488 ms | 0 - 60 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.668 ms | 0 - 83 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.608 ms | 0 - 82 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 15.785 ms | 0 - 31 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 2.26 ms | 0 - 25 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 4.606 ms | 0 - 55 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.361 ms | 0 - 5 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8775P | 2.647 ms | 0 - 59 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8650P | 2.647 ms | 0 - 59 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8255P | 2.647 ms | 0 - 59 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS8450 | 2.608 ms | 0 - 82 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 2.499 ms | 0 - 24 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 12.153 ms | 0 - 38 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 5.059 ms | 0 - 35 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.137 ms | 0 - 61 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA7255P | 4.606 ms | 0 - 55 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8295P | 3.197 ms | 0 - 48 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 1.137 ms | 0 - 61 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.951 ms | 0 - 56 MB | NPU
| MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 5.059 ms | 0 - 35 MB | NPU

## License
* The license for the original implementation of MiniLM-v2 can be found
  [here](https://github.com/UKPLab/sentence-transformers/blob/master/LICENSE).

## References
* [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084)
* [Source Model Implementation](https://github.com/UKPLab/sentence-transformers)

## 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).