Update README.md
Browse files
README.md
CHANGED
|
@@ -33,7 +33,7 @@ This model is a fine-tuned version of **Meta's DINOv3-ViT-S/16** specifically de
|
|
| 33 |
- **License:** Apache 2.0
|
| 34 |
- **Fine-tuned from:** `facebook/dinov3-vits16-pretrain-lvd1689m`
|
| 35 |
|
| 36 |
-
##
|
| 37 |
|
| 38 |
The model was trained using the [openpecha/uchen-ume-classification](https://huggingface.co/datasets/openpecha/uchen-ume-classification) dataset. This training set consists of **4,572 images** balanced across two major categories.
|
| 39 |
|
|
@@ -64,7 +64,7 @@ The binary classes were mapped from the following granular script types:
|
|
| 64 |
|
| 65 |
*Note: Classes labeled "Difficult," "Multi-script," and "Non-Tibetan" were excluded to maintain a clean training signal for the Uchen/Ume boundary.*
|
| 66 |
|
| 67 |
-
##
|
| 68 |
The model achieved its peak performance at **Stage B** (Partial backbone unfreezing of the last 2 blocks).
|
| 69 |
|
| 70 |
- **Test Accuracy:** 98.95%
|
|
@@ -77,7 +77,7 @@ The model achieved its peak performance at **Stage B** (Partial backbone unfreez
|
|
| 77 |
| **Uchen** | 159 | 2 |
|
| 78 |
| **Ume** | 6 | 595 |
|
| 79 |
|
| 80 |
-
##
|
| 81 |
|
| 82 |
```python
|
| 83 |
from transformers import AutoImageProcessor, AutoModelForImageClassification
|
|
|
|
| 33 |
- **License:** Apache 2.0
|
| 34 |
- **Fine-tuned from:** `facebook/dinov3-vits16-pretrain-lvd1689m`
|
| 35 |
|
| 36 |
+
## Dataset & Class Distribution
|
| 37 |
|
| 38 |
The model was trained using the [openpecha/uchen-ume-classification](https://huggingface.co/datasets/openpecha/uchen-ume-classification) dataset. This training set consists of **4,572 images** balanced across two major categories.
|
| 39 |
|
|
|
|
| 64 |
|
| 65 |
*Note: Classes labeled "Difficult," "Multi-script," and "Non-Tibetan" were excluded to maintain a clean training signal for the Uchen/Ume boundary.*
|
| 66 |
|
| 67 |
+
## Performance Summary
|
| 68 |
The model achieved its peak performance at **Stage B** (Partial backbone unfreezing of the last 2 blocks).
|
| 69 |
|
| 70 |
- **Test Accuracy:** 98.95%
|
|
|
|
| 77 |
| **Uchen** | 159 | 2 |
|
| 78 |
| **Ume** | 6 | 595 |
|
| 79 |
|
| 80 |
+
## How to Get Started
|
| 81 |
|
| 82 |
```python
|
| 83 |
from transformers import AutoImageProcessor, AutoModelForImageClassification
|