Update README.md
Browse files
README.md
CHANGED
|
@@ -1,39 +1,115 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 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 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
+
# Welcome to CLIP-as-service!
|
| 5 |
+
|
| 6 |
+
CLIP-as-service is a low-latency high-scalability service for embedding images and text. It can be easily integrated as a microservice into neural search solutions.
|
| 7 |
+
|
| 8 |
+
⚡ Fast: Serve CLIP models with TensorRT, ONNX runtime and PyTorch w/o JIT with 800QPS[*]. Non-blocking duplex streaming on requests and responses, designed for large data and long-running tasks.
|
| 9 |
+
|
| 10 |
+
🫐 Elastic: Horizontally scale up and down multiple CLIP models on single GPU, with automatic load balancing.
|
| 11 |
+
|
| 12 |
+
🐥 Easy-to-use: No learning curve, minimalist design on client and server. Intuitive and consistent API for image and sentence embedding.
|
| 13 |
+
|
| 14 |
+
👒 Modern: Async client support. Easily switch between gRPC, HTTP, WebSocket protocols with TLS and compression.
|
| 15 |
+
|
| 16 |
+
🍱 Integration: Smooth integration with neural search ecosystem including Jina and DocArray. Build cross-modal and multi-modal solutions in no time.
|
| 17 |
+
|
| 18 |
+
[*] with default config (single replica, PyTorch no JIT) on GeForce RTX 3090.
|
| 19 |
+
|
| 20 |
+
## Try it!
|
| 21 |
+
|
| 22 |
+
## Install
|
| 23 |
+
|
| 24 |
+
[PyPI](https://img.shields.io/pypi/v/clip_client?color=%23ffffff&label=%20) is the latest version.
|
| 25 |
+
|
| 26 |
+
Make sure you are using Python 3.7+. You can install the client and server independently. It is **not required** to install both: e.g. you can install `clip_server` on a GPU machine and `clip_client` on a local laptop.
|
| 27 |
+
|
| 28 |
+
Client
|
| 29 |
+
|
| 30 |
+
```bash
|
| 31 |
+
pip install clip-client
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
+
Server (PyTorch)
|
| 35 |
+
|
| 36 |
+
```
|
| 37 |
+
pip install clip-server
|
| 38 |
+
```
|
| 39 |
+
Server (ONNX)
|
| 40 |
+
|
| 41 |
+
```
|
| 42 |
+
pip install "clip_server[onnx]"
|
| 43 |
+
```
|
| 44 |
+
Server (TensorRT)
|
| 45 |
+
|
| 46 |
+
```
|
| 47 |
+
pip install nvidia-pyindex
|
| 48 |
+
pip install "clip_server[tensorrt]"
|
| 49 |
+
```
|
| 50 |
+
Server on [Google Colab](https://colab.research.google.com/github/jina-ai/clip-as-service/blob/main/docs/hosting/cas-on-colab.ipynb)
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
## Quick check
|
| 55 |
+
|
| 56 |
+
After installing, you can run the following commands for a quick connectivity check.
|
| 57 |
+
|
| 58 |
+
### Start the server
|
| 59 |
+
|
| 60 |
+
Start PyTorch Server
|
| 61 |
+
```bash
|
| 62 |
+
python -m clip_server
|
| 63 |
+
```
|
| 64 |
+
Start ONNX Server
|
| 65 |
+
```bash
|
| 66 |
+
python -m clip_server onnx-flow.yml
|
| 67 |
+
```
|
| 68 |
+
Start TensorRT Server
|
| 69 |
+
```bash
|
| 70 |
+
python -m clip_server tensorrt-flow.yml
|
| 71 |
+
```
|
| 72 |
+
|
| 73 |
+
At the first time starting the server, it will download the default pretrained model, which may take a while depending on your network speed. Then you will get the address information similar to the following:
|
| 74 |
+
|
| 75 |
+
```text
|
| 76 |
+
╭────────────── 🔗 Endpoint ───────────────╮
|
| 77 |
+
│ 🔗 Protocol GRPC │
|
| 78 |
+
│ 🏠 Local 0.0.0.0:51000 │
|
| 79 |
+
│ 🔒 Private 192.168.31.62:51000 │
|
| 80 |
+
| 🌍 Public 87.105.159.191:51000 |
|
| 81 |
+
╰──────────────────────────────────────────╯
|
| 82 |
+
```
|
| 83 |
+
|
| 84 |
+
This means the server is ready to serve. Note down the three addresses shown above, you will need them later.
|
| 85 |
+
|
| 86 |
+
### Connect from client
|
| 87 |
+
|
| 88 |
+
```{tip}
|
| 89 |
+
Depending on the location of the client and server. You may use different IP addresses:
|
| 90 |
+
- Client and server are on the same machine: use local address, e.g. `0.0.0.0`
|
| 91 |
+
- Client and server are connected to the same router: use private network address, e.g. `192.168.3.62`
|
| 92 |
+
- Server is in public network: use public network address, e.g. `87.105.159.191`
|
| 93 |
+
```
|
| 94 |
+
|
| 95 |
+
Run the following Python script:
|
| 96 |
+
|
| 97 |
+
```python
|
| 98 |
+
from clip_client import Client
|
| 99 |
+
|
| 100 |
+
c = Client('grpc://0.0.0.0:51000')
|
| 101 |
+
c.profile()
|
| 102 |
+
```
|
| 103 |
+
|
| 104 |
+
will give you:
|
| 105 |
+
|
| 106 |
+
```text
|
| 107 |
+
Roundtrip 16ms 100%
|
| 108 |
+
├── Client-server network 8ms 49%
|
| 109 |
+
└── Server 8ms 51%
|
| 110 |
+
├── Gateway-CLIP network 2ms 25%
|
| 111 |
+
└── CLIP model 6ms 75%
|
| 112 |
+
{'Roundtrip': 15.684750003856607, 'Client-server network': 7.684750003856607, 'Server': 8, 'Gateway-CLIP network': 2, 'CLIP model': 6}
|
| 113 |
+
```
|
| 114 |
+
|
| 115 |
+
It means the client and the server are now connected. Well done!
|