README.md

metadata

library_name: transformers
license: apache-2.0
base_model: facebook/dinov2-base
tags:
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
  - f1
model-index:
  - name: spermatogenesis-classifier
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: imagefolder
          type: imagefolder
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8910256410256411
          - name: F1
            type: f1
            value: 0.8896300082346593

spermatogenesis-classifier

This model is a fine-tuned version of facebook/dinov2-base on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.3150
• Accuracy: 0.8910
• F1: 0.8896
• Acc I-iv: 0.8710
• Acc Ix-x: 0.9048
• Acc V-vi: 0.8511
• Acc Vii-vii: 0.9714
• Acc Xi- xii: 0.8636

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 50
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Acc I-iv	Acc Ix-x	Acc V-vi	Acc Vii-vii	Acc Xi- xii
2.7787	1.0	20	1.2571	0.4295	0.3860	0.0	0.4762	0.5532	0.3143	0.9091
1.6595	2.0	40	1.5081	0.3590	0.3372	0.0323	0.5238	0.0426	0.5714	1.0
1.5573	3.0	60	0.8242	0.5513	0.5636	0.3871	0.5238	0.3830	0.6571	1.0
1.2868	4.0	80	0.7303	0.7885	0.7570	0.5484	0.4762	0.8723	0.9429	1.0
0.9034	5.0	100	0.4915	0.8077	0.8036	0.7097	0.8095	0.7660	0.8857	0.9091
1.2132	6.0	120	0.5243	0.8013	0.7923	0.6452	0.8095	0.9574	0.6571	0.9091
0.8576	7.0	140	0.7115	0.7692	0.7224	0.5806	0.2857	0.8298	1.0	1.0
0.8557	8.0	160	0.5277	0.7692	0.7716	0.8710	0.6667	0.5106	1.0	0.9091
0.7294	9.0	180	0.4170	0.8333	0.8306	0.6129	0.9048	0.8511	0.9143	0.9091
0.6713	10.0	200	0.4585	0.8141	0.8070	0.9032	0.9048	0.7234	0.8571	0.7273
0.7973	11.0	220	0.4767	0.8397	0.8241	0.7097	0.6667	0.8936	0.8857	1.0
0.6637	12.0	240	0.4327	0.8013	0.8057	0.9032	0.7619	0.6170	0.9143	0.9091
0.6284	13.0	260	0.3897	0.8462	0.8331	0.5484	0.8571	0.9149	1.0	0.8636
0.7981	14.0	280	0.3915	0.8654	0.8512	0.6774	0.9524	0.9362	0.9429	0.7727
0.5017	15.0	300	0.3150	0.8910	0.8896	0.8710	0.9048	0.8511	0.9714	0.8636
0.5893	16.0	320	0.3640	0.8526	0.8485	0.8065	0.9048	0.8298	0.9429	0.7727
0.6591	17.0	340	0.3563	0.8718	0.8684	0.7742	0.8571	0.8723	0.9429	0.9091
0.4976	18.0	360	0.3648	0.8397	0.8393	0.9355	0.8095	0.7447	0.8857	0.8636
0.5034	19.0	380	0.3839	0.8462	0.8389	0.6452	0.7619	0.9149	0.9429	0.9091
0.4612	20.0	400	0.3724	0.8654	0.8636	0.9032	0.8571	0.8723	0.8286	0.8636

Framework versions

• Transformers 5.6.2
• Pytorch 2.11.0+cu130
• Datasets 4.8.4
• Tokenizers 0.22.2