Update README.md
Browse files
README.md
CHANGED
|
@@ -88,6 +88,26 @@ model = AutoModelForCausalLM.from_pretrained(
|
|
| 88 |
|
| 89 |
### Training
|
| 90 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
#### Finetuning data:
|
| 92 |
|
| 93 |
- `saillab/alpaca-kurdish_kurmanji-cleaned`
|
|
|
|
| 88 |
|
| 89 |
### Training
|
| 90 |
|
| 91 |
+
Transformers `4.44.2`
|
| 92 |
+
1 NVIDIA A40
|
| 93 |
+
Duration 7h 41m 12s
|
| 94 |
+
|
| 95 |
+
```json
|
| 96 |
+
{
|
| 97 |
+
"total_flos": 2225817933447045000,
|
| 98 |
+
"train/epoch": 0.9998075072184792,
|
| 99 |
+
"train/global_step": 2597,
|
| 100 |
+
"train/grad_norm": 1.172538161277771,
|
| 101 |
+
"train/learning_rate": 0,
|
| 102 |
+
"train/loss": 0.7774,
|
| 103 |
+
"train_loss": 0.892096030377038,
|
| 104 |
+
"train_runtime": 27479.3172,
|
| 105 |
+
"train_samples_per_second": 1.512,
|
| 106 |
+
"train_steps_per_second": 0.095
|
| 107 |
+
}
|
| 108 |
+
```
|
| 109 |
+
|
| 110 |
+
|
| 111 |
#### Finetuning data:
|
| 112 |
|
| 113 |
- `saillab/alpaca-kurdish_kurmanji-cleaned`
|