Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
datasets:
|
| 4 |
+
- isek-ai/danbooru-tags-2016-2023
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
library_name: transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# SDPrompt-RetNet-v2
|
| 11 |
+
|
| 12 |
+
This model is a pretrained RetNet model trained from scratch using https://github.com/syncdoth/RetNet.
|
| 13 |
+
|
| 14 |
+
It achieves the following results on the evaluation set:
|
| 15 |
+
- Loss: 0.5923
|
| 16 |
+
|
| 17 |
+
## Model description
|
| 18 |
+
|
| 19 |
+
More information needed
|
| 20 |
+
|
| 21 |
+
## Intended uses & limitations
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Training and evaluation data
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training procedure
|
| 30 |
+
|
| 31 |
+
### Training hyperparameters
|
| 32 |
+
|
| 33 |
+
The following hyperparameters were used during training:
|
| 34 |
+
- learning_rate: 0.0001
|
| 35 |
+
- train_batch_size: 32
|
| 36 |
+
- eval_batch_size: 8
|
| 37 |
+
- seed: 42
|
| 38 |
+
- gradient_accumulation_steps: 2
|
| 39 |
+
- total_train_batch_size: 64
|
| 40 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
+
- lr_scheduler_type: linear
|
| 42 |
+
- lr_scheduler_warmup_steps: 500
|
| 43 |
+
- num_epochs: 1
|
| 44 |
+
|
| 45 |
+
### Training results
|
| 46 |
+
|
| 47 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 48 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 49 |
+
| 0.975 | 0.07 | 500 | 1.0005 |
|
| 50 |
+
| 0.7549 | 0.13 | 1000 | 0.7604 |
|
| 51 |
+
| 0.6923 | 0.2 | 1500 | 0.7090 |
|
| 52 |
+
| 0.6753 | 0.26 | 2000 | 0.6778 |
|
| 53 |
+
| 0.6591 | 0.33 | 2500 | 0.6568 |
|
| 54 |
+
| 0.6337 | 0.39 | 3000 | 0.6429 |
|
| 55 |
+
| 0.6288 | 0.46 | 3500 | 0.6319 |
|
| 56 |
+
| 0.624 | 0.53 | 4000 | 0.6218 |
|
| 57 |
+
| 0.62 | 0.59 | 4500 | 0.6172 |
|
| 58 |
+
| 0.603 | 0.66 | 5000 | 0.6090 |
|
| 59 |
+
| 0.5931 | 0.72 | 5500 | 0.6032 |
|
| 60 |
+
| 0.5957 | 0.79 | 6000 | 0.5986 |
|
| 61 |
+
| 0.5972 | 0.85 | 6500 | 0.5948 |
|
| 62 |
+
| 0.5928 | 0.92 | 7000 | 0.5926 |
|
| 63 |
+
| 0.5904 | 0.98 | 7500 | 0.5923 |
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
### Framework versions
|
| 67 |
+
|
| 68 |
+
- Transformers 4.36.1
|
| 69 |
+
- Pytorch 2.1.2+cu121
|
| 70 |
+
- Datasets 2.15.0
|
| 71 |
+
- Tokenizers 0.15.0
|