Update README.md
Browse files
README.md
CHANGED
|
@@ -11,6 +11,10 @@ base_model: jimmycarter/LibreFLUX
|
|
| 11 |
|
| 12 |
This model/pipeline is the product of my [LibreFlux ControlNet training repo](https://github.com/NeuralVFX/LibreFLUX-ControlNet), which uses [LibreFLUX](https://huggingface.co/jimmycarter/LibreFLUX) as the underlying Transformer model for the ControlNet. For the dataset, I auto labeled 165K images from the SA1B dataset and trained for 1 epoch. I've tested using this ControlNet model as a base for transfer learning to less generic datasets, the results are good!
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# How does this relate to LibreFLUX?
|
| 15 |
- Base model is [LibreFLUX](https://huggingface.co/jimmycarter/LibreFLUX)
|
| 16 |
- Trained in same non-distilled fashion
|
|
|
|
| 11 |
|
| 12 |
This model/pipeline is the product of my [LibreFlux ControlNet training repo](https://github.com/NeuralVFX/LibreFLUX-ControlNet), which uses [LibreFLUX](https://huggingface.co/jimmycarter/LibreFLUX) as the underlying Transformer model for the ControlNet. For the dataset, I auto labeled 165K images from the SA1B dataset and trained for 1 epoch. I've tested using this ControlNet model as a base for transfer learning to less generic datasets, the results are good!
|
| 13 |
|
| 14 |
+
# FYI
|
| 15 |
+
- The results with the SA1B Dataset suffer from blurry faces, and some other anomalies
|
| 16 |
+
- Planning on retraining using LAION
|
| 17 |
+
|
| 18 |
# How does this relate to LibreFLUX?
|
| 19 |
- Base model is [LibreFLUX](https://huggingface.co/jimmycarter/LibreFLUX)
|
| 20 |
- Trained in same non-distilled fashion
|