outputs_dinov3_448

This model is a fine-tuned version of facebook/dinov3-vitb16-pretrain-lvd1689m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0585
• Precision: 0.9585
• Recall: 0.9716
• F1: 0.9650

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-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.05
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1
No log	1.0	145	0.3163	0.6908	0.6712	0.6809
No log	2.0	290	0.1855	0.8418	0.9223	0.8802
No log	3.0	435	0.1311	0.8984	0.9664	0.9312
0.2777	4.0	580	0.1046	0.9063	0.9548	0.9299
0.2777	5.0	725	0.0824	0.9481	0.9590	0.9535
0.2777	6.0	870	0.0735	0.9316	0.9727	0.9517
0.0778	7.0	1015	0.0657	0.9430	0.9727	0.9576
0.0778	8.0	1160	0.0633	0.9603	0.9643	0.9623
0.0778	9.0	1305	0.0598	0.9595	0.9706	0.9650
0.0778	10.0	1450	0.0585	0.9585	0.9716	0.9650

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2