dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_simkd

This model is a fine-tuned version of WinKawaks/vit-small-patch16-224 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.0864
Accuracy: 0.842
Brier Loss: 0.4773
Nll: 1.3055
F1 Micro: 0.842
F1 Macro: 0.8427
Ece: 0.4404
Aurc: 0.0865

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: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
0.1387	1.0	1000	0.1293	0.4412	0.8613	3.7912	0.4412	0.3863	0.3246	0.3321
0.1154	2.0	2000	0.1132	0.668	0.7441	2.1523	0.668	0.6478	0.4699	0.1467
0.1074	3.0	3000	0.1075	0.73	0.7150	1.9686	0.7300	0.7323	0.5177	0.1104
0.1036	4.0	4000	0.1068	0.7258	0.6943	2.0361	0.7258	0.7299	0.4903	0.1318
0.0996	5.0	5000	0.1047	0.742	0.6897	2.1166	0.7420	0.7456	0.5061	0.1124
0.0967	6.0	6000	0.0992	0.78	0.6307	1.8492	0.78	0.7875	0.5018	0.0924
0.0923	7.0	7000	0.0985	0.774	0.6055	2.0449	0.774	0.7800	0.4698	0.1056
0.0893	8.0	8000	0.0982	0.7817	0.6012	1.9476	0.7817	0.7814	0.4696	0.1243
0.0871	9.0	9000	0.0954	0.8043	0.5826	1.7573	0.8043	0.8065	0.4925	0.0811
0.0857	10.0	10000	0.0969	0.784	0.5672	2.0878	0.7840	0.7846	0.4417	0.1076
0.083	11.0	11000	0.0934	0.8067	0.5658	1.6628	0.8067	0.8062	0.4809	0.0788
0.0819	12.0	12000	0.0930	0.8027	0.5499	1.6718	0.8027	0.8035	0.4592	0.0851
0.081	13.0	13000	0.0937	0.7957	0.5579	1.7282	0.7957	0.7956	0.4544	0.0948
0.079	14.0	14000	0.0939	0.794	0.5463	1.9406	0.7940	0.7974	0.4344	0.1142
0.0779	15.0	15000	0.0912	0.81	0.5206	1.6776	0.81	0.8141	0.4371	0.0947
0.0767	16.0	16000	0.0905	0.813	0.5165	1.6744	0.813	0.8140	0.4383	0.0864
0.0766	17.0	17000	0.0911	0.8113	0.5239	1.7109	0.8113	0.8104	0.4428	0.0902
0.0762	18.0	18000	0.0914	0.8093	0.5153	1.6778	0.8093	0.8098	0.4290	0.0998
0.0759	19.0	19000	0.0904	0.8163	0.5076	1.6946	0.8163	0.8178	0.4333	0.0939
0.075	20.0	20000	0.0897	0.8133	0.5062	1.5892	0.8133	0.8155	0.4300	0.0898
0.0743	21.0	21000	0.0895	0.8147	0.5058	1.5900	0.8148	0.8149	0.4315	0.0917
0.0745	22.0	22000	0.0898	0.8157	0.5014	1.5523	0.8157	0.8164	0.4287	0.0848
0.0737	23.0	23000	0.0901	0.8127	0.5038	1.6625	0.8128	0.8146	0.4219	0.0978
0.0735	24.0	24000	0.0907	0.8117	0.5082	1.6475	0.8117	0.8133	0.4231	0.1064
0.0732	25.0	25000	0.0901	0.8103	0.5041	1.6830	0.8103	0.8105	0.4187	0.1017
0.0727	26.0	26000	0.0899	0.8135	0.5015	1.6499	0.8135	0.8170	0.4197	0.1020
0.0722	27.0	27000	0.0880	0.8265	0.4931	1.4651	0.8265	0.8273	0.4330	0.0975
0.0718	28.0	28000	0.0876	0.8263	0.4917	1.4213	0.8263	0.8275	0.4354	0.0858
0.0725	29.0	29000	0.0891	0.8247	0.4930	1.5581	0.8247	0.8254	0.4288	0.0946
0.0717	30.0	30000	0.0879	0.8327	0.4913	1.4417	0.8327	0.8326	0.4403	0.0888
0.0715	31.0	31000	0.0872	0.8375	0.4866	1.3775	0.8375	0.8389	0.4435	0.0872
0.0715	32.0	32000	0.0884	0.8297	0.4915	1.5136	0.8297	0.8305	0.4331	0.0946
0.0717	33.0	33000	0.0877	0.8347	0.4851	1.4096	0.8347	0.8347	0.4375	0.0845
0.0716	34.0	34000	0.0880	0.8323	0.4866	1.4547	0.8323	0.8333	0.4323	0.0926
0.0713	35.0	35000	0.0873	0.8343	0.4833	1.3884	0.8343	0.8351	0.4375	0.0810
0.0713	36.0	36000	0.0873	0.8365	0.4843	1.4168	0.8365	0.8372	0.4381	0.0913
0.071	37.0	37000	0.0871	0.8393	0.4831	1.3524	0.8393	0.8399	0.4412	0.0882
0.0709	38.0	38000	0.0877	0.834	0.4862	1.4457	0.834	0.8353	0.4371	0.0929
0.071	39.0	39000	0.0870	0.836	0.4811	1.3954	0.836	0.8367	0.4360	0.0886
0.0708	40.0	40000	0.0867	0.8387	0.4800	1.3687	0.8387	0.8403	0.4390	0.0867
0.0706	41.0	41000	0.0866	0.8395	0.4802	1.3464	0.8395	0.8399	0.4412	0.0860
0.0708	42.0	42000	0.0868	0.8363	0.4796	1.3828	0.8363	0.8371	0.4345	0.0886
0.0709	43.0	43000	0.0866	0.838	0.4790	1.3503	0.838	0.8390	0.4382	0.0860
0.0702	44.0	44000	0.0866	0.8415	0.4787	1.3679	0.8415	0.8425	0.4403	0.0899
0.0703	45.0	45000	0.0866	0.8373	0.4788	1.3192	0.8373	0.8379	0.4374	0.0863
0.0702	46.0	46000	0.0865	0.841	0.4776	1.3357	0.841	0.8417	0.4398	0.0871
0.0703	47.0	47000	0.0864	0.8417	0.4772	1.3302	0.8417	0.8424	0.4406	0.0859
0.0705	48.0	48000	0.0865	0.841	0.4776	1.3096	0.841	0.8417	0.4398	0.0877
0.0703	49.0	49000	0.0864	0.8413	0.4775	1.3022	0.8413	0.8419	0.4400	0.0865
0.0703	50.0	50000	0.0864	0.842	0.4773	1.3055	0.842	0.8427	0.4404	0.0865

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_simkd

dit-base-finetuned-rvlcdip-small_rvl_cdip-NK1000_simkd

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions