dinov2-s-201

This model is a fine-tuned version of facebook/dinov2-small-imagenet1k-1-layer on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5503
• Accuracy: 0.8049

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	5	1.7244	0.2195
1.4057	2.0	10	1.1285	0.5122
1.4057	3.0	15	0.6513	0.7561
0.8392	4.0	20	0.5946	0.8049
0.8392	5.0	25	0.6221	0.8293
0.6571	6.0	30	1.3668	0.4878
0.6571	7.0	35	0.6909	0.6585
0.7314	8.0	40	0.6185	0.7073
0.7314	9.0	45	1.1204	0.5122
0.6679	10.0	50	0.6920	0.7073
0.6679	11.0	55	0.5515	0.7561
0.5023	12.0	60	0.8328	0.6829
0.5023	13.0	65	0.5849	0.7805
0.5507	14.0	70	0.4574	0.8293
0.5507	15.0	75	0.7229	0.7317
0.4605	16.0	80	0.6463	0.6829
0.4605	17.0	85	0.5158	0.7805
0.3592	18.0	90	0.5429	0.7317
0.3592	19.0	95	0.4544	0.8293
0.3719	20.0	100	0.5683	0.7805
0.3719	21.0	105	0.7423	0.7073
0.4792	22.0	110	0.6053	0.7561
0.4792	23.0	115	0.5218	0.8049
0.3421	24.0	120	0.5553	0.8049
0.3421	25.0	125	0.6367	0.7805
0.3528	26.0	130	0.3843	0.8049
0.3528	27.0	135	0.6923	0.7317
0.3335	28.0	140	0.6799	0.7073
0.3335	29.0	145	1.0437	0.6098
0.2933	30.0	150	0.8362	0.7073
0.2933	31.0	155	0.6174	0.7073
0.2902	32.0	160	0.5487	0.8780
0.2902	33.0	165	0.6631	0.8049
0.3046	34.0	170	0.7015	0.7561
0.3046	35.0	175	0.5250	0.8049
0.2355	36.0	180	0.6684	0.8537
0.2355	37.0	185	0.5820	0.7805
0.21	38.0	190	0.7903	0.7805
0.21	39.0	195	0.4358	0.9024
0.1833	40.0	200	0.8039	0.8293
0.1833	41.0	205	0.6242	0.8537
0.2227	42.0	210	0.7574	0.7073
0.2227	43.0	215	0.8873	0.7561
0.1831	44.0	220	0.9501	0.7561
0.1831	45.0	225	0.8774	0.8293
0.1815	46.0	230	0.7826	0.8049
0.1815	47.0	235	1.1516	0.6829
0.1615	48.0	240	0.6514	0.8537
0.1615	49.0	245	0.5799	0.8049
0.1381	50.0	250	0.7545	0.7805
0.1381	51.0	255	0.5452	0.8049
0.1462	52.0	260	0.7610	0.8049
0.1462	53.0	265	0.7827	0.8049
0.1096	54.0	270	0.6393	0.8537
0.1096	55.0	275	0.5902	0.8293
0.0914	56.0	280	0.7998	0.8537
0.0914	57.0	285	0.9032	0.7805
0.1674	58.0	290	0.5467	0.8537
0.1674	59.0	295	0.9872	0.7805
0.086	60.0	300	0.6481	0.8537

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1