https://github.com/ArmanZarei/Mitigating-T2I-Comp-Issues

https://github.com/ArmanZarei/Mitigating-T2I-Comp-Issues

Last synced: 2 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/ArmanZarei/Mitigating-T2I-Comp-Issues
• Owner: ArmanZarei
• Created: 2024-06-10T10:09:34.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-09-16T18:46:56.000Z (4 months ago)
• Last Synced: 2024-09-16T23:43:42.626Z (4 months ago)
• Language: Python
• Size: 22.5 KB
• Stars: 6
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-diffusion-categorized - [Code

README

        
# Mitigating-T2I-Compositionality-Issues

Training WiCLP sample script

```python

python train_text_emb_projection.py \

    --output_dir="final_experiments/window_based_linear_projection/sd2_1/color" \

    --stable_diffusion_checkpoint="stabilityai/stable-diffusion-2-1" \

    --image_size=768 \

    --evaluation_batch_size=10 \

    --train_batch_size=4 \

    --projector_type="window_aware_linear" \

    --projection_window_size=5 \

    --train_steps=25000 \

    --validation_steps=1000 \

    --wandb_run_name="Color dataset - SD v2.1"

```