dinov2-base-imagenet1k-1-layer-head-finetuned-galaxy_mnist

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

• Loss: 0.4438
• Accuracy: 0.8385
• Precision: 0.8388
• Recall: 0.8385
• F1: 0.8386

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: 0.0001
• 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
1.5682	0.99	31	1.4744	0.279	0.3518	0.279	0.2447
1.1895	1.98	62	1.1500	0.5175	0.5367	0.5175	0.5208
0.8876	2.98	93	0.8869	0.673	0.6822	0.673	0.6692
0.7366	4.0	125	0.7267	0.744	0.7472	0.744	0.7433
0.6499	4.99	156	0.6573	0.753	0.7566	0.753	0.7533
0.6147	5.98	187	0.6209	0.7695	0.7760	0.7695	0.7695
0.5911	6.98	218	0.5802	0.7855	0.7869	0.7855	0.7860
0.5489	8.0	250	0.5573	0.799	0.8019	0.799	0.7992
0.5306	8.99	281	0.5318	0.801	0.8005	0.801	0.8004
0.5154	9.98	312	0.5177	0.8095	0.8098	0.8095	0.8096
0.4963	10.98	343	0.5097	0.8155	0.8157	0.8155	0.8149
0.5304	12.0	375	0.4971	0.814	0.8144	0.814	0.8139
0.4779	12.99	406	0.4889	0.817	0.8182	0.817	0.8160
0.4774	13.98	437	0.4822	0.8245	0.8255	0.8245	0.8250
0.4852	14.98	468	0.4772	0.827	0.8290	0.827	0.8272
0.4693	16.0	500	0.4724	0.8285	0.8308	0.8285	0.8290
0.456	16.99	531	0.4636	0.829	0.8294	0.829	0.8289
0.4703	17.98	562	0.4623	0.829	0.8303	0.829	0.8291
0.4715	18.98	593	0.4609	0.8305	0.8329	0.8305	0.8308
0.4775	20.0	625	0.4574	0.835	0.8369	0.835	0.8353
0.4776	20.99	656	0.4531	0.833	0.8338	0.833	0.8332
0.4599	21.98	687	0.4500	0.8325	0.8333	0.8325	0.8325
0.4391	22.98	718	0.4510	0.832	0.8341	0.832	0.8321
0.4537	24.0	750	0.4466	0.835	0.8356	0.835	0.8352
0.4555	24.99	781	0.4451	0.8355	0.8365	0.8355	0.8356
0.4476	25.98	812	0.4438	0.8385	0.8388	0.8385	0.8386
0.4618	26.98	843	0.4440	0.8355	0.8366	0.8355	0.8358
0.4734	28.0	875	0.4432	0.835	0.8362	0.835	0.8352
0.4453	28.99	906	0.4433	0.8365	0.8378	0.8365	0.8367
0.4591	29.76	930	0.4430	0.8365	0.8376	0.8365	0.8367

Framework versions

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