dinov2-base-imagenet1k-1-layer-finetuned-galaxy10-decals-finetuned-galaxy_mnist

This model is a fine-tuned version of matthieulel/dinov2-base-imagenet1k-1-layer-finetuned-galaxy10-decals on the matthieulel/galaxy_mnist dataset. It achieves the following results on the evaluation set:

• Loss: 0.1834
• Accuracy: 0.943
• Precision: 0.9430
• Recall: 0.943
• F1: 0.9430

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: 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
0.8616	0.99	31	0.4898	0.8345	0.8394	0.8345	0.8342
0.3162	1.98	62	0.1963	0.921	0.9217	0.921	0.9210
0.2596	2.98	93	0.1666	0.9315	0.9322	0.9315	0.9313
0.2237	4.0	125	0.1579	0.9385	0.9386	0.9385	0.9385
0.2304	4.99	156	0.1631	0.936	0.9364	0.936	0.9361
0.2096	5.98	187	0.1686	0.933	0.9341	0.933	0.9329
0.1935	6.98	218	0.1660	0.934	0.9341	0.934	0.9339
0.1829	8.0	250	0.1596	0.9415	0.9418	0.9415	0.9415
0.178	8.99	281	0.1613	0.937	0.9381	0.937	0.9370
0.158	9.98	312	0.1697	0.9335	0.9358	0.9335	0.9334
0.1767	10.98	343	0.1653	0.935	0.9350	0.935	0.9349
0.176	12.0	375	0.1752	0.936	0.9375	0.936	0.9357
0.1563	12.99	406	0.1892	0.932	0.9339	0.932	0.9319
0.1499	13.98	437	0.1946	0.9345	0.9353	0.9345	0.9344
0.1388	14.98	468	0.1763	0.937	0.9371	0.937	0.9370
0.1418	16.0	500	0.1875	0.9375	0.9390	0.9375	0.9375
0.1332	16.99	531	0.1769	0.9365	0.9364	0.9365	0.9364
0.1413	17.98	562	0.1851	0.9355	0.9363	0.9355	0.9355
0.1363	18.98	593	0.1834	0.943	0.9430	0.943	0.9430
0.1454	20.0	625	0.1823	0.938	0.9384	0.938	0.9380
0.1369	20.99	656	0.1834	0.938	0.9380	0.938	0.9380
0.1298	21.98	687	0.1960	0.932	0.9335	0.932	0.9318
0.1215	22.98	718	0.1756	0.941	0.9410	0.941	0.9410
0.1206	24.0	750	0.1917	0.9395	0.9397	0.9395	0.9394
0.1173	24.99	781	0.1873	0.937	0.9370	0.937	0.9370
0.1071	25.98	812	0.1856	0.9375	0.9376	0.9375	0.9375
0.1259	26.98	843	0.1871	0.938	0.9380	0.938	0.9380
0.1245	28.0	875	0.1866	0.9395	0.9396	0.9395	0.9395
0.1065	28.99	906	0.1870	0.94	0.9401	0.94	0.9400
0.1066	29.76	930	0.1862	0.941	0.9410	0.941	0.9410

Framework versions

• Transformers 4.37.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.15.1