Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -18,7 +18,41 @@ metrics:
|
|
| 18 |
- M贸dulos: q_proj, k_proj, v_proj
|
| 19 |
- 4-bit NF4
|
| 20 |
- Early Stopping: patience=3
|
| 21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
## Uso
|
| 23 |
```python
|
| 24 |
from peft import PeftModel
|
|
|
|
| 18 |
- M贸dulos: q_proj, k_proj, v_proj
|
| 19 |
- 4-bit NF4
|
| 20 |
- Early Stopping: patience=3
|
| 21 |
+
## Entrenamiento
|
| 22 |
+
Training logs (manual, Epoch estimado):
|
| 23 |
+
| Step | Epoch | Training Loss | Validation Loss |
|
| 24 |
+
|---|---|---|---|
|
| 25 |
+
| 100 | 0.046 | 0.820900 | 0.792622 |
|
| 26 |
+
| 200 | 0.093 | 0.770500 | 0.764106 |
|
| 27 |
+
| 300 | 0.139 | 0.762600 | 0.754589 |
|
| 28 |
+
| 400 | 0.186 | 0.733300 | 0.741709 |
|
| 29 |
+
| 500 | 0.232 | 0.734900 | 0.735551 |
|
| 30 |
+
| 600 | 0.279 | 0.741500 | 0.731295 |
|
| 31 |
+
| 700 | 0.325 | 0.722700 | 0.710327 |
|
| 32 |
+
| 800 | 0.371 | 0.735200 | 0.703414 |
|
| 33 |
+
| 900 | 0.418 | 0.721500 | 0.693650 |
|
| 34 |
+
| 1000 | 0.464 | 0.697900 | 0.690272 |
|
| 35 |
+
| 1100 | 0.511 | 0.689100 | 0.684814 |
|
| 36 |
+
| 1200 | 0.557 | 0.662200 | 0.674680 |
|
| 37 |
+
| 1300 | 0.604 | 0.664400 | 0.677307 |
|
| 38 |
+
| 1400 | 0.650 | 0.663100 | 0.669781 |
|
| 39 |
+
| 1500 | 0.696 | 0.616000 | 0.665949 |
|
| 40 |
+
| 1600 | 0.743 | 0.622500 | 0.664927 |
|
| 41 |
+
| 1700 | 0.789 | 0.622200 | 0.658744 |
|
| 42 |
+
| 1800 | 0.836 | 0.630300 | 0.654155 |
|
| 43 |
+
| 1900 | 0.882 | 0.628300 | 0.656066 |
|
| 44 |
+
| 2000 | 0.929 | 0.612600 | 0.653236 |
|
| 45 |
+
| 2100 | 0.975 | 0.619600 | 0.647662 |
|
| 46 |
+
| 2200 | 1.021 | 0.605400 | 0.649643 |
|
| 47 |
+
| 2300 | 1.068 | 0.603700 | 0.646184 |
|
| 48 |
+
| 2400 | 1.114 | 0.600100 | 0.643537 |
|
| 49 |
+
| 2500 | 1.161 | 0.565200 | 0.642405 |
|
| 50 |
+
| 2600 | 1.207 | 0.594800 | 0.636302 |
|
| 51 |
+
| 2700 | 1.253 | 0.587300 | 0.630301 |
|
| 52 |
+
| 2800 | 1.300 | 0.598400 | 0.628895 |
|
| 53 |
+
| 2900 | 1.346 | 0.561300 | 0.630126 |
|
| 54 |
+
| 3000 | 1.393 | 0.538800 | 0.633145 |
|
| 55 |
+
| 3100 | 1.439 | 0.537100 | 0.632617 |
|
| 56 |
## Uso
|
| 57 |
```python
|
| 58 |
from peft import PeftModel
|