dinov2-base-imagenet1k-1-layer-head-finetuned-100-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: 0.4478
• Accuracy: 0.8335
• Precision: 0.8342
• Recall: 0.8335
• F1: 0.8337

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
1.5876	0.99	31	1.5424	0.257	0.2412	0.257	0.2234
1.2929	1.98	62	1.2250	0.4485	0.4644	0.4485	0.4223
0.9596	2.98	93	0.9329	0.677	0.6826	0.677	0.6683
0.7585	4.0	125	0.7491	0.759	0.7590	0.759	0.7590
0.6636	4.99	156	0.6671	0.78	0.7816	0.78	0.7797
0.621	5.98	187	0.6224	0.7895	0.7923	0.7895	0.7889
0.6004	6.98	218	0.5860	0.7895	0.7916	0.7895	0.7901
0.5454	8.0	250	0.5620	0.797	0.8009	0.797	0.7969
0.5357	8.99	281	0.5372	0.804	0.8045	0.804	0.8039
0.5137	9.98	312	0.5223	0.805	0.8061	0.805	0.8050
0.4968	10.98	343	0.5123	0.812	0.8125	0.812	0.8122
0.5295	12.0	375	0.5009	0.8165	0.8182	0.8165	0.8169
0.4882	12.99	406	0.4921	0.8185	0.8197	0.8185	0.8178
0.4839	13.98	437	0.4868	0.817	0.8184	0.817	0.8175
0.4857	14.98	468	0.4819	0.818	0.8207	0.818	0.8182
0.4692	16.0	500	0.4781	0.8205	0.8233	0.8205	0.8210
0.4559	16.99	531	0.4689	0.8265	0.8276	0.8265	0.8265
0.4689	17.98	562	0.4675	0.825	0.8267	0.825	0.8251
0.4695	18.98	593	0.4666	0.82	0.8229	0.82	0.8204
0.4772	20.0	625	0.4631	0.821	0.8238	0.821	0.8214
0.4757	20.99	656	0.4571	0.8315	0.8323	0.8315	0.8318
0.4633	21.98	687	0.4537	0.832	0.8324	0.832	0.8320
0.4407	22.98	718	0.4547	0.826	0.8285	0.826	0.8261
0.4525	24.0	750	0.4508	0.831	0.8319	0.831	0.8313
0.4556	24.99	781	0.4494	0.8305	0.8317	0.8305	0.8307
0.4468	25.98	812	0.4478	0.8335	0.8342	0.8335	0.8337
0.4579	26.98	843	0.4481	0.8325	0.8336	0.8325	0.8328
0.4749	28.0	875	0.4472	0.832	0.8331	0.832	0.8323
0.4427	28.99	906	0.4472	0.8325	0.8337	0.8325	0.8328
0.4652	29.76	930	0.4468	0.8325	0.8336	0.8325	0.8328

Framework versions

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