https://github.com/thoth000/embeddingedit

The diffusion model for editing word embedding space
https://github.com/thoth000/embeddingedit

diffusion embeddings pytorch text2image

Last synced: about 2 months ago
JSON representation

The diffusion model for editing word embedding space

• Host: GitHub
• URL: https://github.com/thoth000/embeddingedit
• Owner: thoth000
• Created: 2025-04-19T05:03:23.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2025-05-17T11:28:24.000Z (about 1 year ago)
• Last Synced: 2025-05-17T11:31:44.368Z (about 1 year ago)
• Topics: diffusion, embeddings, pytorch, text2image
• Language: Python
• Homepage:
• Size: 5.94 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# EmbeddingEdit

![sample image](https://github.com/thoth000/EmbeddingEdit/blob/main/docs/images/image.png)

[**Website**](https://thoth000.github.io/EmbeddingEdit/)

## Features

- I implement each step of image generation, so interpolation is possible at any stage.

  1. prompt tokenization

  2. raw word embedding

  3. prompt embedding

  4. denozing

- image generation from a prompt

- images generation along with { `linear`, ... }interpolation between { `prompt`, } embedding

## Tutorial

1. You have to prepare the environment.

   ```bash

   pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

   pip install diffusers["torch"] transformers

   pip install accelerate

   ```

2. You can run programs.

   - You are available to generate image from a prompt.

     ```bash

     python single.py --prompt "a dog playing with ball"

     ```

   - You are also available to generate few images along with linear interpolation.

     ```bash

     python interp.py --prompt_a "Only an apple" --prompt_b "ten apples"

     ```