Commit
·
e9a8379
1
Parent(s):
7b6edac
update model card README.md
Browse files
README.md
CHANGED
|
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [microsoft/swinv2-tiny-patch4-window8-256](https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256) on an unknown dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
-
- Loss:
|
| 20 |
-
- Accuracy: 0.
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
@@ -45,42 +45,17 @@ The following hyperparameters were used during training:
|
|
| 45 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
- lr_scheduler_type: linear
|
| 47 |
- lr_scheduler_warmup_ratio: 0.1
|
| 48 |
-
- num_epochs:
|
| 49 |
|
| 50 |
### Training results
|
| 51 |
|
| 52 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|
| 53 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
|
| 54 |
-
| No log | 1.0 | 3 | 1.
|
| 55 |
-
| No log | 2.0 | 6 | 1.
|
| 56 |
-
| No log | 3.0 | 9 | 1.
|
| 57 |
-
| 1.
|
| 58 |
-
| 1.
|
| 59 |
-
| 1.5014 | 6.0 | 18 | 1.1866 | 0.5 |
|
| 60 |
-
| 1.2791 | 7.0 | 21 | 1.1289 | 0.4762 |
|
| 61 |
-
| 1.2791 | 8.0 | 24 | 1.0491 | 0.5 |
|
| 62 |
-
| 1.2791 | 9.0 | 27 | 0.9757 | 0.5476 |
|
| 63 |
-
| 1.1172 | 10.0 | 30 | 0.9237 | 0.6190 |
|
| 64 |
-
| 1.1172 | 11.0 | 33 | 0.8831 | 0.7143 |
|
| 65 |
-
| 1.1172 | 12.0 | 36 | 0.8543 | 0.6667 |
|
| 66 |
-
| 1.1172 | 13.0 | 39 | 0.8334 | 0.6905 |
|
| 67 |
-
| 0.9564 | 14.0 | 42 | 0.8143 | 0.7381 |
|
| 68 |
-
| 0.9564 | 15.0 | 45 | 0.8189 | 0.7857 |
|
| 69 |
-
| 0.9564 | 16.0 | 48 | 0.7855 | 0.8095 |
|
| 70 |
-
| 0.8806 | 17.0 | 51 | 0.7927 | 0.7619 |
|
| 71 |
-
| 0.8806 | 18.0 | 54 | 0.8064 | 0.7143 |
|
| 72 |
-
| 0.8806 | 19.0 | 57 | 0.7396 | 0.7143 |
|
| 73 |
-
| 0.8616 | 20.0 | 60 | 0.7181 | 0.7857 |
|
| 74 |
-
| 0.8616 | 21.0 | 63 | 0.7331 | 0.7381 |
|
| 75 |
-
| 0.8616 | 22.0 | 66 | 0.7079 | 0.7619 |
|
| 76 |
-
| 0.8616 | 23.0 | 69 | 0.6732 | 0.8095 |
|
| 77 |
-
| 0.808 | 24.0 | 72 | 0.6667 | 0.8095 |
|
| 78 |
-
| 0.808 | 25.0 | 75 | 0.6749 | 0.7857 |
|
| 79 |
-
| 0.808 | 26.0 | 78 | 0.7134 | 0.6905 |
|
| 80 |
-
| 0.8247 | 27.0 | 81 | 0.7134 | 0.7143 |
|
| 81 |
-
| 0.8247 | 28.0 | 84 | 0.6892 | 0.7857 |
|
| 82 |
-
| 0.8247 | 29.0 | 87 | 0.6769 | 0.7857 |
|
| 83 |
-
| 0.8127 | 30.0 | 90 | 0.6746 | 0.7619 |
|
| 84 |
|
| 85 |
|
| 86 |
### Framework versions
|
|
|
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [microsoft/swinv2-tiny-patch4-window8-256](https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256) on an unknown dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 1.2462
|
| 20 |
+
- Accuracy: 0.4762
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 45 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
- lr_scheduler_type: linear
|
| 47 |
- lr_scheduler_warmup_ratio: 0.1
|
| 48 |
+
- num_epochs: 5
|
| 49 |
|
| 50 |
### Training results
|
| 51 |
|
| 52 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|
| 53 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
|
| 54 |
+
| No log | 1.0 | 3 | 1.4504 | 0.3095 |
|
| 55 |
+
| No log | 2.0 | 6 | 1.3403 | 0.4048 |
|
| 56 |
+
| No log | 3.0 | 9 | 1.2864 | 0.5238 |
|
| 57 |
+
| 1.459 | 4.0 | 12 | 1.2617 | 0.4762 |
|
| 58 |
+
| 1.459 | 5.0 | 15 | 1.2462 | 0.4762 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
|
| 61 |
### Framework versions
|