[AI Cog] Want to Run an AI Business but Lack a GPU? Cant Set Up the Environment? Use Cog to Easily Deploy Your Business to the Cloud

When you want to start an AI business but lack a GPU, what should you do?

Consider using Cog to deploy AI services to the cloud, serverless.

Let’s see how to use Cog to get it on the cloud.

Find a development server.

Cog

Installation

1
2


sudo curl -o /usr/local/bin/cog -L https://github.com/replicate/cog/releases/latest/download/cog_`uname -s`_`uname -m`
sudo chmod +x /usr/local/bin/cog

Verification

This step is optional, mainly to verify if your environment is okay.

1

sudo cog predict r8.im/stability-ai/stable-diffusion@sha256:f178fa7a1ae43a9a9af01b833b9d2ecf97b1bcb0acfd2dc5dd04895e042863f1 -i prompt="a pot of gold"

Initialization

1

cog init

Generate the main files.

1
2


├── cog.yaml # Similar to Docker file, defining the environment
├── predict.py # Inference code

Writing Code

Modify the code as follows.

cog.yaml is similar to a Docker file, defining the environment.

 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


# Configuration for Cog ⚙️
# Reference: https://cog.run/yaml

build:
  # set to true if your model requires a GPU
  gpu: false

  # a list of ubuntu apt packages to install
  # system_packages:
  #   - "libgl1-mesa-glx"
  #   - "libglib2.0-0"

  # python version in the form '3.11' or '3.11.4'
  python_version: "3.10"

  # a list of packages in the format <package-name>==<version>
  # python_packages:
  #   - "numpy==1.19.4"
  #   - "torch==1.8.0"
  #   - "torchvision==0.9.0"

  # commands run after the environment is setup
  # run:
  #   - "echo env is ready!"
  #   - "echo another command if needed"

# predict.py defines how predictions are run on your model
predict: "predict.py:Predictor"

predict.py defines the inputs (name: str, scale: float), outputs (str), and the inference process.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23


# Prediction interface for Cog ⚙️
# https://cog.run/python

from cog import BasePredictor, Input, Path

class Predictor(BasePredictor):
    def setup(self) -> None:
        """Load the model into memory to make running multiple predictions efficient"""
        # self.model = torch.load("./weights.pth")

    def predict(
        self,
        name: str = Input(description="Your name"),
        # image: Path = Input(description="Grayscale input image"),
        scale: float = Input(
            description="Factor to scale image by", ge=0, le=10, default=1.5
        ),
    ) -> str:
        """Run a single prediction on the model"""
        # processed_input = preprocess(image)
        # output = self.model(processed_image, scale)
        # return postprocess(output)
        return "hello " + name + " and scale " + str(scale)

Local Testing

Test it out.

1

cog predict -i name=Learn AI from Scratch

Output.

1
2
3


Starting Docker image cog-git-base and running setup()...
Running prediction...
hello Learn AI from Scratch and scale 1.5

Deployment

On the cloud create model

Push the model to the cloud.

1
2


cog login
cog push r8.im/<your-username>/<your-model-name>

Cloud Testing

cog-input

cog-output

Test successful!

Afterwards, you can call it via API.

Conclusion

This article demonstrates the entire process of using Cog to deploy to the cloud.

The example does not use a GPU. Check the documentation if needed.