https://github.com/jmaczan/text-to-ml

Homebrew AutoML using natural text. Like HuggingGPT + LangChain + type inference
https://github.com/jmaczan/text-to-ml

automl deep-learning huggingface hugginggpt machine-learning python

Last synced: 2 months ago
JSON representation

Homebrew AutoML using natural text. Like HuggingGPT + LangChain + type inference

• Host: GitHub
• URL: https://github.com/jmaczan/text-to-ml
• Owner: jmaczan
• License: gpl-3.0
• Created: 2024-04-04T12:32:21.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2024-06-04T08:36:13.000Z (8 months ago)
• Last Synced: 2024-06-04T09:55:40.525Z (8 months ago)
• Topics: automl, deep-learning, huggingface, hugginggpt, machine-learning, python
• Language: Python
• Homepage: https://pagedout.institute/download/PagedOut_004_beta1.pdf#page=4
• Size: 242 KB
• Stars: 7
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# 🎮 text-to-ml

> 💜 PagedOut #4 Issue ["Building automated machine learning with type inference"](https://pagedout.institute/download/PagedOut_004_beta1.pdf#page=4)

Run AutoML using natural text. Like HuggingGPT + LangChain + type inference

> 🏭 A breakdown of what is going on in code you can read on my blog: [maczan.pl](https://maczan.pl/p/lets-build-text-to-ml-an-automl-library)

It picks a right model from Hugging Face library based on user natural language query and then runs the model and parses the output to a type, inferred from the query



> ⚡ You can run this code in [Lightning AI Studio template](https://lightning.ai/jed/studios/build-your-own-automl-using-hugging-face-inference-client-and-openai-api)





  





It's still an early project and **you are welcome to contribute**!

## Setup

1. Get OpenAI API Key

2. Get Hugging Face API Key

3. Create an assistant and copy its id

4. Create `.env` file and fill it with values:

```

OPENAI_API_KEY=

HF_TOKEN=

```

## Build

```sh

conda create -n text-to-ml python=3.9

conda activate text-to-ml

conda install --file requirements.txt

```

## Run

```

python app.py

```

## Run experiments

```

python experiments.py

```

## Cite

If you use this software in your research, please use the following citation:

```bibtex

@misc{Maczan_TextToML_2024,

  title = "Programmable automated machine learning - proof of concept",

  author = "{Maczan, Jędrzej Paweł}",

  howpublished = "\url{https://github.com/jmaczan/text-to-ml}",

  year = 2024,

  publisher = {GitHub}

}

```

## License

GPLv3

## Author

Jędrzej Paweł Maczan, Poland, 2024