Hugging Face's logo

sofa566
/
git-base-pokemon

Image-Text-to-Text
Transformers
PyTorch
TensorBoard
git
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
git-base-pokemon / README.md
sofa566's picture
sofa566
update model card README.md
a9b055d
preview code
|
raw
history blame contribute delete
2.82 kB
metadata
license: mit
tags:
  - generated_from_trainer
datasets:
  - imagefolder
model-index:
  - name: git-base-pokemon
    results: []

git-base-pokemon

This model is a fine-tuned version of microsoft/git-base on the imagefolder dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0371
  • Wer Score: 2.4731

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: 5e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Score
7.3268 2.13 50 4.4847 21.4974
2.2464 4.26 100 0.3519 11.4118
0.1049 6.38 150 0.0302 0.7468
0.0223 8.51 200 0.0270 0.4668
0.0137 10.64 250 0.0280 3.5742
0.0073 12.77 300 0.0304 7.1240
0.0034 14.89 350 0.0309 6.4885
0.0018 17.02 400 0.0326 5.0499
0.0011 19.15 450 0.0335 5.2302
0.0009 21.28 500 0.0342 4.3645
0.0007 23.4 550 0.0346 5.1445
0.0006 25.53 600 0.0351 4.0639
0.0006 27.66 650 0.0355 3.8862
0.0006 29.79 700 0.0359 3.4514
0.0006 31.91 750 0.0363 3.0486
0.0006 34.04 800 0.0363 2.8645
0.0006 36.17 850 0.0366 2.7199
0.0006 38.3 900 0.0369 2.6675
0.0006 40.43 950 0.0369 2.6304
0.0006 42.55 1000 0.0370 2.4910
0.0006 44.68 1050 0.0370 2.4834
0.0006 46.81 1100 0.0371 2.4629
0.0006 48.94 1150 0.0371 2.4731

Framework versions

  • Transformers 4.29.1
  • Pytorch 1.12.1
  • Datasets 2.11.0
  • Tokenizers 0.11.0