vit-base-patch32-224-in21k-finetuned-galaxy10-decals

This model is a fine-tuned version of google/vit-base-patch32-224-in21k on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

• Loss: 0.5055
• Accuracy: 0.8422
• Precision: 0.8413
• Recall: 0.8422
• F1: 0.8406

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: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• 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
2.0954	0.99	31	1.9514	0.3737	0.3322	0.3737	0.2452
1.4835	1.98	62	1.3284	0.6184	0.6016	0.6184	0.5878
1.1252	2.98	93	0.9771	0.7300	0.7400	0.7300	0.7004
0.9605	4.0	125	0.8374	0.7570	0.7754	0.7570	0.7368
0.8383	4.99	156	0.7286	0.7762	0.7728	0.7762	0.7650
0.7665	5.98	187	0.7256	0.7689	0.7683	0.7689	0.7586
0.7305	6.98	218	0.6640	0.7948	0.8031	0.7948	0.7966
0.6689	8.0	250	0.6792	0.7807	0.7859	0.7807	0.7708
0.6783	8.99	281	0.5985	0.8117	0.8076	0.8117	0.8071
0.6225	9.98	312	0.6118	0.8050	0.8036	0.8050	0.8025
0.6081	10.98	343	0.5966	0.8112	0.8108	0.8112	0.8080
0.6028	12.0	375	0.5708	0.8202	0.8239	0.8202	0.8199
0.6052	12.99	406	0.6035	0.8010	0.8116	0.8010	0.7982
0.5553	13.98	437	0.5542	0.8196	0.8199	0.8196	0.8143
0.5526	14.98	468	0.5385	0.8326	0.8346	0.8326	0.8317
0.5199	16.0	500	0.5298	0.8219	0.8192	0.8219	0.8172
0.4974	16.99	531	0.5291	0.8298	0.8306	0.8298	0.8260
0.5015	17.98	562	0.5244	0.8275	0.8280	0.8275	0.8267
0.4763	18.98	593	0.5190	0.8354	0.8357	0.8354	0.8316
0.4763	20.0	625	0.5241	0.8264	0.8286	0.8264	0.8249
0.4592	20.99	656	0.5061	0.8410	0.8439	0.8410	0.8406
0.4414	21.98	687	0.5207	0.8269	0.8265	0.8269	0.8260
0.4372	22.98	718	0.5342	0.8253	0.8283	0.8253	0.8254
0.4118	24.0	750	0.5256	0.8275	0.8291	0.8275	0.8274
0.4319	24.99	781	0.5055	0.8422	0.8413	0.8422	0.8406
0.3807	25.98	812	0.5187	0.8377	0.8375	0.8377	0.8361
0.4066	26.98	843	0.5203	0.8348	0.8333	0.8348	0.8326
0.376	28.0	875	0.5128	0.8365	0.8361	0.8365	0.8348
0.3992	28.99	906	0.5108	0.8377	0.8375	0.8377	0.8364
0.3743	29.76	930	0.5087	0.8388	0.8389	0.8388	0.8378

Framework versions

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