InstaGPT

Description

This is an experimental automated instagram manager, showcasing capabilities of the large language model ChatGPT and text-to-image models like Dall-E and Stable Diffusion for content creation. The instagram account @fromage_fiesta showcases the possibility of the project to fully automate the generation of posts, following a content and schedule plan.

Note: Currently the text-to-image engine uses OpenAI's Dall-E. In the next version, it is possible to choose between the engines Dall-E, Stable Diffusion and Midjourney.

Installation

This project uses the dependency management and packaging system Poetry. Make sure that Python is installed and Poetry is configured correctly. The project requires Python 3.9 or newer, if another version is needed, please adapt ./pyproject.toml. Alternatively to Poetry you can use your own virtual environment and install the required packages as usual through requirements.txt.

Setup

1. Get an OpenAI API Key. Optional: To use Stable Diffusion for image generation, you additionaly need to get a Replicate API Key.

2. Clone the repository.

3. Create a new Python environment:

> poetry env use <path to python executable>

> poetry shell

4. Install the necessary dependencies from pyproject.toml:

> poetry install

5. Build package:

> poetry build

Configuration

1. Find the file named .env.template in the main folder. This file may be hidden by default in some operating systems due to the dot prefix.
2. Create a copy of .env.template and call it .env; if you're already in a command prompt/terminal window:

> cp .env.template .env.

3. Open the .env file in a text editor.
4. Find the line that says OPENAI_API_KEY=.
5. After the =, enter your unique OpenAI API Key without any quotes or spaces.
6. Find the lines that say USER_ID= and PASSWORD=.
7. After the =, enter your unique Instagram credentials without any quotes or spaces.
8. (Optional) Find the line that says REPLICATE_API_TOKEN=.
9. (Optional) After the =, enter your unique Replicate API Key without any quotes or spaces.
10. Save and close the .env file.

Usage

Configurations

1. Create a csv file with the topics you want to be posted and save it in the directory. The csv should follow the format:

topic,posted
blueberry,False
strawberry,False
raspberry,False

2. Adapt the prompt design for your needs. The prompt templates for caption and scene generation are located at insta_gpt/prompt_design. Sources for best practices in prompt engineering are for example reddit and openai docs.
3. Adapt the config.yaml file with the filename of the csv, openai related configurations and if you want to use dalle or stable_diffusion as a model for image generation. Currently, it is best practice to use the "gpt-3.5-turbo" model, because of the cost and performance. If an increase in performance is required, "gpt-4" model can also be set. Note that the pricing of "gpt-4" is by far higher.

Run

You can run the script main.py using Python:

> poetry run python main.py

If operating on a Unix system, the script can also be scheduled to run with cronjobs by adding an entry to the crontab file.

> sudo crontab -u $(whoami) -e

Details on schedule expressions can be found on crontab.guru. Cronjob does need absolute paths. To get the path of python in poetry, run the command poetry run which python. Here is an example of a crontab file running everyday at 9:00 AM:

# Run the script every day at 9:00 AM
0 9 * * * /path/to/poetry/python /path/to/insta_gpt/main.py >> /path/to/insta_gpt/cron.log 2>&1

To end a cronjob, edit the crontab file or kill the cronjob:

> crontab -r

Disclaimer

Please note that the use of a GPT language model and text-to-image models can be expensive due to its token usage. By utilizing this project, you acknowledge that you are responsible for monitoring and managing your own token usage and the associated costs. It is highly recommended to check your OpenAI API and Replicate usage regularly and set up any necessary limits or alerts to prevent unexpected charges.