Update README.md
Browse files
README.md
CHANGED
@@ -32,3 +32,20 @@ https://github.com/hiyouga/LLaMA-Factory
|
|
32 |
| LLama3 | 49.91 | 66.62|
|
33 |
| LLama3-Instruct | 50.55 | 67.15|
|
34 |
| LLama3-Instruct-sft-ruozhiba-3epoch | 50.87 | 67.51|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
| LLama3 | 49.91 | 66.62|
|
33 |
| LLama3-Instruct | 50.55 | 67.15|
|
34 |
| LLama3-Instruct-sft-ruozhiba-3epoch | 50.87 | 67.51|
|
35 |
+
|
36 |
+
### Training hyperparameters
|
37 |
+
The following hyperparameters were used during training:
|
38 |
+
- learning_rate: 5e-05
|
39 |
+
- train_batch_size: 1
|
40 |
+
- eval_batch_size: 1
|
41 |
+
- seed: 42
|
42 |
+
- distributed_type: multi-GPU
|
43 |
+
- num_devices: 8
|
44 |
+
- gradient_accumulation_steps: 2
|
45 |
+
- total_train_batch_size: 16
|
46 |
+
- total_eval_batch_size: 8
|
47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
48 |
+
- lr_scheduler_type: cosine
|
49 |
+
- lr_scheduler_warmup_steps: 20
|
50 |
+
- num_epochs: 3.0
|
51 |
+
- mixed_precision_training: Native AMP
|