update README.md file
Browse files
README.md
CHANGED
|
@@ -128,27 +128,7 @@ FALLBACK class strengthened for production safety.
|
|
| 128 |
|
| 129 |
---
|
| 130 |
|
| 131 |
-
## 6.
|
| 132 |
-
|
| 133 |
-
Base Model: MuRIL (Multilingual BERT for Indic languages)
|
| 134 |
-
|
| 135 |
-
Hyperparameters:
|
| 136 |
-
|
| 137 |
-
* Epochs: 5
|
| 138 |
-
* Batch Size: 16 (with gradient accumulation = 2)
|
| 139 |
-
* Learning Rate: 5e-5
|
| 140 |
-
* Scheduler: Cosine
|
| 141 |
-
* Weight Decay: 0.01
|
| 142 |
-
* Early Stopping Enabled
|
| 143 |
-
* Weighted Cross Entropy Loss
|
| 144 |
-
|
| 145 |
-
Max Sequence Length: 64
|
| 146 |
-
|
| 147 |
-
Hardware: GPU (CUDA)
|
| 148 |
-
|
| 149 |
-
---
|
| 150 |
-
|
| 151 |
-
## 7. Evaluation Results
|
| 152 |
|
| 153 |
### Overall Performance (Test Set: 22,173 samples)
|
| 154 |
|
|
@@ -177,7 +157,7 @@ Hardware: GPU (CUDA)
|
|
| 177 |
|
| 178 |
---
|
| 179 |
|
| 180 |
-
# Example Usage
|
| 181 |
|
| 182 |
```python
|
| 183 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
|
|
|
| 128 |
|
| 129 |
---
|
| 130 |
|
| 131 |
+
## 6. Evaluation Results
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 132 |
|
| 133 |
### Overall Performance (Test Set: 22,173 samples)
|
| 134 |
|
|
|
|
| 157 |
|
| 158 |
---
|
| 159 |
|
| 160 |
+
## 7. Example Usage
|
| 161 |
|
| 162 |
```python
|
| 163 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|