https://github.com/andreasjansson/AutoCog
https://github.com/andreasjansson/AutoCog
Last synced: 14 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/andreasjansson/AutoCog
- Owner: andreasjansson
- Created: 2023-04-15T12:58:54.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2024-09-30T19:42:33.000Z (8 months ago)
- Last Synced: 2025-04-30T21:52:05.254Z (17 days ago)
- Language: Smarty
- Size: 6.19 MB
- Stars: 39
- Watchers: 4
- Forks: 7
- Open Issues: 8
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-replicate - Autocog - Generate predict.py and cog.yaml automatically using GPT-4. (Open-source tools)
README
# AutoCog
[](https://badge.fury.io/py/autocog)
_Generate [predict.py and cog.yaml](https://github.com/replicate/cog) automatically using GPT4_
## Install
```
pip install autocog
```
## Usage
First, specify your AI provider with `--ai-provider` and set your OpenAI/Anthropic API key in an environment variable
```
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-...
```
In the repo you want to cog-ify, run
```
autocog
```
This will generate a cog.yaml and predict.py based on the files in the current directory. It will then run the model and if it fails to run, it will attempt to fix the error and run it again. By default it has 5 attempts to fix it, which can be changed with the `--attempts` flag.
If your model needs a GPU to run, you need to run AutoCog on a GPU machine.
### Human in the loop
Sometimes AutoCog fails to create a working Cog configuration. In those cases you, the human, have to step in and edit the cog.yaml and predict.py files.
Once you have edited them, let AutoCog continue by running `autocog` again. If you'd like to recreate `predict.py` and `cog.yaml` from scratch, run `autocog --initialize`.
By default, AutoCog will guess a `cog predict` command to run the model. If you want to specify your own predict command, use the `--predict-command` flag.
If you want AutoCog to take the generation of `cog.yaml` and `predict.py` in a specific direction, you can use the `--tell` flag to prompt GPT4:
```
autocog --tell="Add inputs to allow for inpainting"
```