metadata
license: mit
base_model: microsoft/phi-2
tags:
- generated_from_trainer
model-index:
- name: V0414H1
results: []
V0414H1
This model is a fine-tuned version of microsoft/phi-2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0480
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine_with_restarts
- lr_scheduler_warmup_steps: 60
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
1.9763 | 0.09 | 10 | 0.8092 |
0.2747 | 0.18 | 20 | 0.1471 |
0.1297 | 0.27 | 30 | 0.1053 |
0.1068 | 0.36 | 40 | 0.0913 |
0.089 | 0.45 | 50 | 0.0842 |
0.0931 | 0.54 | 60 | 0.0789 |
0.0813 | 0.63 | 70 | 0.0783 |
0.0754 | 0.73 | 80 | 0.0770 |
0.0815 | 0.82 | 90 | 0.0686 |
0.0755 | 0.91 | 100 | 0.0683 |
0.0827 | 1.0 | 110 | 0.0709 |
0.0722 | 1.09 | 120 | 0.0680 |
0.0704 | 1.18 | 130 | 0.0583 |
0.0638 | 1.27 | 140 | 0.0583 |
0.0526 | 1.36 | 150 | 0.0541 |
0.0543 | 1.45 | 160 | 0.0515 |
0.0507 | 1.54 | 170 | 0.0500 |
0.0492 | 1.63 | 180 | 0.0490 |
0.05 | 1.72 | 190 | 0.0487 |
0.0527 | 1.81 | 200 | 0.0479 |
0.043 | 1.9 | 210 | 0.0479 |
0.0486 | 1.99 | 220 | 0.0480 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1