Edit model card

convnextv2-tiny-1k-224-finetuned-galaxy10-decals

This model is a fine-tuned version of facebook/convnextv2-tiny-1k-224 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3993
  • Accuracy: 0.8732
  • Precision: 0.8714
  • Recall: 0.8732
  • F1: 0.8715

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: 64
  • eval_batch_size: 64
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 256
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 30

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1
1.8139 0.99 62 1.6803 0.4628 0.4589 0.4628 0.3836
1.0894 2.0 125 0.9304 0.6984 0.6965 0.6984 0.6800
0.8423 2.99 187 0.6630 0.7880 0.7858 0.7880 0.7826
0.6564 4.0 250 0.5769 0.8055 0.8091 0.8055 0.7970
0.5927 4.99 312 0.5283 0.8241 0.8276 0.8241 0.8240
0.5853 6.0 375 0.5106 0.8303 0.8342 0.8303 0.8237
0.5757 6.99 437 0.4490 0.8540 0.8514 0.8540 0.8521
0.5235 8.0 500 0.4651 0.8546 0.8578 0.8546 0.8536
0.5166 8.99 562 0.4501 0.8563 0.8551 0.8563 0.8523
0.486 10.0 625 0.4352 0.8647 0.8624 0.8647 0.8626
0.4882 10.99 687 0.4296 0.8613 0.8594 0.8613 0.8597
0.4426 12.0 750 0.4314 0.8579 0.8614 0.8579 0.8566
0.457 12.99 812 0.4226 0.8641 0.8642 0.8641 0.8624
0.4512 14.0 875 0.4319 0.8619 0.8653 0.8619 0.8591
0.4059 14.99 937 0.4124 0.8692 0.8675 0.8692 0.8681
0.4147 16.0 1000 0.3993 0.8732 0.8714 0.8732 0.8715
0.3721 16.99 1062 0.4116 0.8636 0.8609 0.8636 0.8604
0.3908 18.0 1125 0.4098 0.8675 0.8663 0.8675 0.8665
0.3836 18.99 1187 0.4188 0.8670 0.8651 0.8670 0.8651
0.3716 20.0 1250 0.4172 0.8681 0.8653 0.8681 0.8661
0.3484 20.99 1312 0.4404 0.8653 0.8649 0.8653 0.8628
0.3895 22.0 1375 0.4194 0.8698 0.8689 0.8698 0.8688
0.3452 22.99 1437 0.4447 0.8630 0.8634 0.8630 0.8621
0.341 24.0 1500 0.4253 0.8720 0.8722 0.8720 0.8712
0.3481 24.99 1562 0.4325 0.8681 0.8656 0.8681 0.8658
0.3115 26.0 1625 0.4340 0.8619 0.8609 0.8619 0.8603
0.313 26.99 1687 0.4329 0.8653 0.8644 0.8653 0.8644
0.3362 28.0 1750 0.4329 0.8653 0.8636 0.8653 0.8639
0.3056 28.99 1812 0.4342 0.8658 0.8645 0.8658 0.8644
0.3206 29.76 1860 0.4343 0.8664 0.8648 0.8664 0.8649

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.3.0
  • Datasets 2.19.1
  • Tokenizers 0.15.1
Downloads last month
8
Safetensors
Model size
27.9M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for matthieulel/convnextv2-tiny-1k-224-finetuned-galaxy10-decals

Finetuned
(37)
this model