swinv2-base-patch4-window8-256-finetuned-galaxy10-decals

This model is a fine-tuned version of microsoft/swinv2-base-patch4-window8-256 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

• Loss: 0.4829
• Accuracy: 0.8540
• Precision: 0.8529
• Recall: 0.8540
• F1: 0.8520

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: 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
1.6195	0.99	62	1.4006	0.5101	0.4910	0.5101	0.4782
0.9423	2.0	125	0.7209	0.7616	0.7617	0.7616	0.7531
0.8171	2.99	187	0.5842	0.8010	0.7950	0.8010	0.7938
0.6609	4.0	250	0.5000	0.8224	0.8159	0.8224	0.8143
0.5927	4.99	312	0.5367	0.8191	0.8211	0.8191	0.8184
0.624	6.0	375	0.4946	0.8286	0.8295	0.8286	0.8212
0.5891	6.99	437	0.5068	0.8219	0.8244	0.8219	0.8201
0.5597	8.0	500	0.5071	0.8230	0.8382	0.8230	0.8198
0.5292	8.99	562	0.4464	0.8444	0.8462	0.8444	0.8426
0.5143	10.0	625	0.4556	0.8371	0.8420	0.8371	0.8350
0.5122	10.99	687	0.4765	0.8382	0.8433	0.8382	0.8369
0.4647	12.0	750	0.4900	0.8365	0.8443	0.8365	0.8348
0.4769	12.99	812	0.4639	0.8427	0.8475	0.8427	0.8396
0.4804	14.0	875	0.4468	0.8484	0.8499	0.8484	0.8461
0.4452	14.99	937	0.4492	0.8512	0.8522	0.8512	0.8505
0.4283	16.0	1000	0.4660	0.8433	0.8446	0.8433	0.8401
0.3788	16.99	1062	0.4689	0.8478	0.8454	0.8478	0.8444
0.41	18.0	1125	0.4543	0.8506	0.8502	0.8506	0.8480
0.4007	18.99	1187	0.4766	0.8478	0.8511	0.8478	0.8455
0.406	20.0	1250	0.4716	0.8478	0.8474	0.8478	0.8444
0.3777	20.99	1312	0.5026	0.8455	0.8454	0.8455	0.8430
0.3972	22.0	1375	0.5108	0.8393	0.8402	0.8393	0.8371
0.3665	22.99	1437	0.4934	0.8489	0.8498	0.8489	0.8474
0.3569	24.0	1500	0.4989	0.8495	0.8495	0.8495	0.8478
0.3735	24.99	1562	0.4918	0.8495	0.8468	0.8495	0.8468
0.3301	26.0	1625	0.4927	0.8512	0.8512	0.8512	0.8488
0.3438	26.99	1687	0.4829	0.8540	0.8529	0.8540	0.8520
0.3553	28.0	1750	0.4935	0.8540	0.8530	0.8540	0.8512
0.3312	28.99	1812	0.4882	0.8517	0.8509	0.8517	0.8491
0.3319	29.76	1860	0.4876	0.8517	0.8516	0.8517	0.8497

Framework versions

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