Update README.md
Browse files
README.md
CHANGED
|
@@ -49,8 +49,8 @@ The good news is that the model shows clear evidence of being undertrained, with
|
|
| 49 |
```
|
| 50 |
Training Progress (Micro vs Macro F1):
|
| 51 |
Epoch 2: Micro-F1: 0.595 Macro-F1: 0.226
|
| 52 |
-
Epoch 3: Micro-F1: 0.606 Macro-F1: 0.268
|
| 53 |
-
Epoch 3.5: Micro-F1: 0.611 Macro-F1: 0.290
|
| 54 |
```
|
| 55 |
This makes sense as 3.5 epochs really isn't alot of training time at all.
|
| 56 |
|
|
|
|
| 49 |
```
|
| 50 |
Training Progress (Micro vs Macro F1):
|
| 51 |
Epoch 2: Micro-F1: 0.595 Macro-F1: 0.226
|
| 52 |
+
Epoch 3: Micro-F1: 0.606 Macro-F1: 0.268 (+4.2%)
|
| 53 |
+
Epoch 3.5: Micro-F1: 0.611 Macro-F1: 0.290 (+2.2% only 0.5 epochs)
|
| 54 |
```
|
| 55 |
This makes sense as 3.5 epochs really isn't alot of training time at all.
|
| 56 |
|