Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,7 @@ language:
|
|
| 3 |
- en
|
| 4 |
base_model:
|
| 5 |
- Qwen/Qwen-Image
|
|
|
|
| 6 |
pipeline_tag: text-to-image
|
| 7 |
tags:
|
| 8 |
- '360'
|
|
@@ -50,7 +51,10 @@ Training was first performed using nf4 quantization for 32 epochs (8 epochs coun
|
|
| 50 |
Training then continued at int8 quantization for another 16 epochs (4 epochs counting the original + augmentations as a single epoch):
|
| 51 |
- `qwen-360-diffusion-int8-bf16-v1.safetensors` was trained for a total of 48 epochs (2,304,000 steps)
|
| 52 |
|
| 53 |
-
Training
|
|
|
|
|
|
|
|
|
|
| 54 |
|
| 55 |
---
|
| 56 |
|
|
|
|
| 3 |
- en
|
| 4 |
base_model:
|
| 5 |
- Qwen/Qwen-Image
|
| 6 |
+
- Qwen/Qwen-Image-2512
|
| 7 |
pipeline_tag: text-to-image
|
| 8 |
tags:
|
| 9 |
- '360'
|
|
|
|
| 51 |
Training then continued at int8 quantization for another 16 epochs (4 epochs counting the original + augmentations as a single epoch):
|
| 52 |
- `qwen-360-diffusion-int8-bf16-v1.safetensors` was trained for a total of 48 epochs (2,304,000 steps)
|
| 53 |
|
| 54 |
+
Training continued at int8 quantization for another x epochs on the [Qwen/Qwen-Image-2512](https://huggingface.co/Qwen/Qwen-Image-2512) base model:
|
| 55 |
+
- The total number of equirectangular images increased to 35k, and regularization images were removed from the dataset.
|
| 56 |
+
- The loss was adjusted to be based on the information density of equirectangular images.
|
| 57 |
+
- `qwen-360-diffusion-2512-int8-bf16-v2.safetensors` was trained for x epochs
|
| 58 |
|
| 59 |
---
|
| 60 |
|