metadata
license: mit
base_model: microsoft/phi-2
tags:
- generated_from_trainer
model-index:
- name: V0316MP1
results: []
V0316MP1
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: 1.5025
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: 0.0003
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 32
- 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: 20
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.4218 | 0.09 | 10 | 2.3701 |
| 2.3588 | 0.17 | 20 | 2.3216 |
| 2.2547 | 0.26 | 30 | 2.2504 |
| 2.0897 | 0.34 | 40 | 2.1789 |
| 1.9766 | 0.43 | 50 | 2.1106 |
| 1.8207 | 0.51 | 60 | 2.0495 |
| 1.7309 | 0.6 | 70 | 2.0001 |
| 1.666 | 0.68 | 80 | 1.9488 |
| 1.5586 | 0.77 | 90 | 1.9120 |
| 1.4977 | 0.85 | 100 | 1.8712 |
| 1.422 | 0.94 | 110 | 1.8324 |
| 1.3569 | 1.02 | 120 | 1.7940 |
| 1.2811 | 1.11 | 130 | 1.7640 |
| 1.2312 | 1.19 | 140 | 1.7329 |
| 1.1463 | 1.28 | 150 | 1.7065 |
| 1.1087 | 1.37 | 160 | 1.6802 |
| 1.0139 | 1.45 | 170 | 1.6581 |
| 0.968 | 1.54 | 180 | 1.6377 |
| 0.9078 | 1.62 | 190 | 1.6183 |
| 0.871 | 1.71 | 200 | 1.6013 |
| 0.8252 | 1.79 | 210 | 1.5863 |
| 0.7983 | 1.88 | 220 | 1.5675 |
| 0.7561 | 1.96 | 230 | 1.5566 |
| 0.7413 | 2.05 | 240 | 1.5443 |
| 0.7156 | 2.13 | 250 | 1.5348 |
| 0.701 | 2.22 | 260 | 1.5243 |
| 0.673 | 2.3 | 270 | 1.5174 |
| 0.6627 | 2.39 | 280 | 1.5126 |
| 0.648 | 2.47 | 290 | 1.5119 |
| 0.6553 | 2.56 | 300 | 1.5088 |
| 0.6447 | 2.65 | 310 | 1.5051 |
| 0.6227 | 2.73 | 320 | 1.5045 |
| 0.6338 | 2.82 | 330 | 1.5023 |
| 0.6224 | 2.9 | 340 | 1.5017 |
| 0.6115 | 2.99 | 350 | 1.5025 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1