V0424MADP2 / README.md
L0I6T1Z9Y
upload V0424MADP2
ab71298
metadata
license: mit
base_model: microsoft/phi-2
tags:
  - generated_from_trainer
model-index:
  - name: V0424MADP2
    results: []

V0424MADP2

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: 0.1460

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: 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: 3
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
8.4064 0.09 10 2.9572
4.2383 0.18 20 1.6263
0.8815 0.27 30 0.4726
0.2149 0.36 40 0.2592
0.1734 0.45 50 0.1721
0.16 0.54 60 0.1630
0.161 0.63 70 0.1882
0.1616 0.73 80 0.1665
0.1626 0.82 90 0.1612
0.1634 0.91 100 0.1572
0.1599 1.0 110 0.1501
0.1522 1.09 120 0.1523
0.1575 1.18 130 0.1518
0.1502 1.27 140 0.1513
0.154 1.36 150 0.1491
0.151 1.45 160 0.1499
0.1537 1.54 170 0.1536
0.1524 1.63 180 0.1511
0.1532 1.72 190 0.1545
0.1531 1.81 200 0.1490
0.1577 1.9 210 0.1494
0.1519 1.99 220 0.1519
0.1548 2.08 230 0.1493
0.1462 2.18 240 0.1474
0.1471 2.27 250 0.1474
0.1496 2.36 260 0.1486
0.1483 2.45 270 0.1466
0.1468 2.54 280 0.1473
0.1468 2.63 290 0.1463
0.148 2.72 300 0.1464
0.1472 2.81 310 0.1461
0.1484 2.9 320 0.1460
0.1494 2.99 330 0.1460

Framework versions

  • Transformers 4.36.0.dev0
  • Pytorch 2.1.2+cu121
  • Datasets 2.18.0
  • Tokenizers 0.14.1