vit-base-patch16-224-in21k-finetuned-galaxy10-decals

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:

Loss: 0.4968
Accuracy: 0.8495
Precision: 0.8482
Recall: 0.8495
F1: 0.8470

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
2.0234	0.99	31	1.8634	0.4352	0.5730	0.4352	0.3241
1.3451	1.98	62	1.1766	0.7029	0.6679	0.7029	0.6757
0.9736	2.98	93	0.8081	0.7886	0.7841	0.7886	0.7669
0.8332	4.0	125	0.7286	0.7897	0.7922	0.7897	0.7750
0.7341	4.99	156	0.6676	0.8095	0.8187	0.8095	0.8004
0.651	5.98	187	0.6159	0.8168	0.8227	0.8168	0.8122
0.6181	6.98	218	0.5935	0.8224	0.8262	0.8224	0.8216
0.5731	8.0	250	0.5548	0.8258	0.8289	0.8258	0.8228
0.5794	8.99	281	0.5410	0.8292	0.8275	0.8292	0.8270
0.5154	9.98	312	0.5438	0.8264	0.8268	0.8264	0.8236
0.5063	10.98	343	0.5247	0.8320	0.8326	0.8320	0.8308
0.496	12.0	375	0.5067	0.8377	0.8404	0.8377	0.8374
0.4775	12.99	406	0.5067	0.8393	0.8392	0.8393	0.8375
0.4456	13.98	437	0.5007	0.8433	0.8436	0.8433	0.8392
0.4463	14.98	468	0.4824	0.8489	0.8528	0.8489	0.8469
0.4191	16.0	500	0.4900	0.8455	0.8459	0.8455	0.8401
0.3904	16.99	531	0.5002	0.8439	0.8463	0.8439	0.8409
0.3833	17.98	562	0.5225	0.8382	0.8480	0.8382	0.8397
0.3717	18.98	593	0.4762	0.8489	0.8491	0.8489	0.8472
0.3664	20.0	625	0.4968	0.8495	0.8482	0.8495	0.8470
0.3463	20.99	656	0.5061	0.8433	0.8458	0.8433	0.8403
0.3324	21.98	687	0.4962	0.8472	0.8461	0.8472	0.8454
0.3277	22.98	718	0.5116	0.8450	0.8449	0.8450	0.8436
0.3037	24.0	750	0.5043	0.8489	0.8508	0.8489	0.8486
0.3124	24.99	781	0.5108	0.8427	0.8433	0.8427	0.8422
0.2831	25.98	812	0.5170	0.8388	0.8404	0.8388	0.8384
0.3043	26.98	843	0.5151	0.8450	0.8456	0.8450	0.8445
0.284	28.0	875	0.5072	0.8450	0.8433	0.8450	0.8429
0.2849	28.99	906	0.5151	0.8439	0.8431	0.8439	0.8425
0.2797	29.76	930	0.5114	0.8439	0.8429	0.8439	0.8423