Edit model card

convnextv2-nano-22k-224-finetuned-galaxy10-decals

This model is a fine-tuned version of facebook/convnextv2-nano-22k-224 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4386
  • Accuracy: 0.8687
  • Precision: 0.8680
  • Recall: 0.8687
  • F1: 0.8662

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
1.7173 0.99 62 1.5497 0.4617 0.4305 0.4617 0.4119
0.9692 2.0 125 0.8180 0.7306 0.7295 0.7306 0.7246
0.7643 2.99 187 0.6075 0.7931 0.7921 0.7931 0.7880
0.6282 4.0 250 0.5506 0.8151 0.8112 0.8151 0.8068
0.5712 4.99 312 0.5769 0.7982 0.8129 0.7982 0.8002
0.5702 6.0 375 0.5195 0.8315 0.8351 0.8315 0.8225
0.5423 6.99 437 0.4890 0.8331 0.8296 0.8331 0.8303
0.4989 8.0 500 0.4764 0.8371 0.8361 0.8371 0.8342
0.4997 8.99 562 0.4725 0.8405 0.8393 0.8405 0.8365
0.476 10.0 625 0.4582 0.8467 0.8465 0.8467 0.8435
0.4603 10.99 687 0.4460 0.8489 0.8464 0.8489 0.8472
0.4318 12.0 750 0.4398 0.8534 0.8519 0.8534 0.8515
0.4387 12.99 812 0.4575 0.8613 0.8598 0.8613 0.8577
0.4357 14.0 875 0.4398 0.8568 0.8541 0.8568 0.8532
0.3944 14.99 937 0.4425 0.8540 0.8533 0.8540 0.8524
0.3961 16.0 1000 0.4394 0.8574 0.8555 0.8574 0.8542
0.3557 16.99 1062 0.4510 0.8523 0.8497 0.8523 0.8481
0.3881 18.0 1125 0.4399 0.8591 0.8590 0.8591 0.8577
0.3663 18.99 1187 0.4631 0.8546 0.8545 0.8546 0.8524
0.3691 20.0 1250 0.4439 0.8608 0.8585 0.8608 0.8577
0.3443 20.99 1312 0.4524 0.8568 0.8555 0.8568 0.8545
0.3728 22.0 1375 0.4386 0.8687 0.8680 0.8687 0.8662
0.3309 22.99 1437 0.4506 0.8585 0.8578 0.8585 0.8573
0.33 24.0 1500 0.4426 0.8630 0.8613 0.8630 0.8618
0.3541 24.99 1562 0.4625 0.8585 0.8561 0.8585 0.8560
0.2968 26.0 1625 0.4460 0.8613 0.8593 0.8613 0.8590
0.3031 26.99 1687 0.4492 0.8641 0.8630 0.8641 0.8628
0.3207 28.0 1750 0.4480 0.8664 0.8640 0.8664 0.8637
0.2949 28.99 1812 0.4478 0.8636 0.8614 0.8636 0.8615
0.2985 29.76 1860 0.4477 0.8636 0.8612 0.8636 0.8614

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.3.0
  • Datasets 2.19.1
  • Tokenizers 0.15.1
Downloads last month
0
Safetensors
Model size
15M 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-nano-22k-224-finetuned-galaxy10-decals

Finetuned
(1)
this model