vit-base-patch32-384-finetuned-galaxy10-decals

This model is a fine-tuned version of google/vit-base-patch32-384 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

• Loss: 0.5542
• Accuracy: 0.8326
• Precision: 0.8324
• Recall: 0.8326
• F1: 0.8298

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
1.68	0.99	31	1.3835	0.5259	0.5014	0.5259	0.4922
0.9395	1.98	62	0.8286	0.7120	0.7053	0.7120	0.6986
0.7814	2.98	93	0.7194	0.7604	0.7515	0.7604	0.7456
0.7227	4.0	125	0.6271	0.7818	0.7913	0.7818	0.7743
0.6309	4.99	156	0.5944	0.7959	0.7959	0.7959	0.7952
0.5754	5.98	187	0.5448	0.8112	0.8165	0.8112	0.8087
0.5519	6.98	218	0.5456	0.8010	0.7990	0.8010	0.7991
0.5077	8.0	250	0.5458	0.8191	0.8229	0.8191	0.8160
0.5086	8.99	281	0.5326	0.8174	0.8181	0.8174	0.8146
0.455	9.98	312	0.5379	0.8174	0.8179	0.8174	0.8143
0.4532	10.98	343	0.5239	0.8247	0.8238	0.8247	0.8225
0.4311	12.0	375	0.5290	0.8202	0.8197	0.8202	0.8169
0.4399	12.99	406	0.5355	0.8236	0.8269	0.8236	0.8213
0.4026	13.98	437	0.5132	0.8303	0.8288	0.8303	0.8268
0.3964	14.98	468	0.5101	0.8269	0.8290	0.8269	0.8247
0.3649	16.0	500	0.5296	0.8253	0.8242	0.8253	0.8222
0.3353	16.99	531	0.5319	0.8236	0.8212	0.8236	0.8198
0.3372	17.98	562	0.5203	0.8303	0.8315	0.8303	0.8300
0.3281	18.98	593	0.5428	0.8315	0.8319	0.8315	0.8289
0.3152	20.0	625	0.5453	0.8264	0.8283	0.8264	0.8262
0.3016	20.99	656	0.5464	0.8224	0.8252	0.8224	0.8192
0.2826	21.98	687	0.5473	0.8241	0.8214	0.8241	0.8213
0.2832	22.98	718	0.5596	0.8275	0.8281	0.8275	0.8255
0.2547	24.0	750	0.5768	0.8247	0.8260	0.8247	0.8243
0.2682	24.99	781	0.5693	0.8230	0.8244	0.8230	0.8226
0.245	25.98	812	0.5542	0.8326	0.8324	0.8326	0.8298
0.2575	26.98	843	0.5665	0.8241	0.8254	0.8241	0.8234
0.2386	28.0	875	0.5716	0.8309	0.8314	0.8309	0.8293
0.2452	28.99	906	0.5659	0.8303	0.8295	0.8303	0.8279
0.2394	29.76	930	0.5674	0.8315	0.8313	0.8315	0.8294

Framework versions

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