cdip-small_rvl_cdip-NK1000_kd_test

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: 0.3813
Accuracy: 0.8558
Brier Loss: 0.2176
Nll: 1.4251
F1 Micro: 0.8558
F1 Macro: 0.8566
Ece: 0.0597
Aurc: 0.0299

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: 24
eval_batch_size: 24
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
1.7536	1.0	667	0.9652	0.6695	0.4474	2.2965	0.6695	0.6595	0.0494	0.1257
0.8802	2.0	1334	0.7683	0.7195	0.3806	2.0303	0.7195	0.7116	0.0473	0.0920
0.5767	3.0	2001	0.6276	0.7698	0.3253	1.9446	0.7698	0.7711	0.0436	0.0684
0.4263	4.0	2668	0.6095	0.7785	0.3110	1.9810	0.7785	0.7810	0.0474	0.0624
0.3987	5.0	3335	0.5608	0.791	0.2939	1.8539	0.791	0.7918	0.0504	0.0557
0.3179	6.0	4002	0.6057	0.7935	0.3027	1.8778	0.7935	0.7940	0.0811	0.0548
0.2428	7.0	4669	0.5828	0.8043	0.2905	1.8616	0.8043	0.8050	0.0662	0.0520
0.2094	8.0	5336	0.5812	0.7957	0.2973	1.8459	0.7957	0.8019	0.0783	0.0532
0.1715	9.0	6003	0.6152	0.7987	0.2993	1.9533	0.7987	0.7998	0.0723	0.0539
0.1508	10.0	6670	0.5442	0.808	0.2820	1.8159	0.808	0.8097	0.0836	0.0476
0.1434	11.0	7337	0.4881	0.828	0.2549	1.6938	0.828	0.8286	0.0610	0.0410
0.1267	12.0	8004	0.4720	0.8365	0.2465	1.6878	0.8365	0.8360	0.0576	0.0400
0.115	13.0	8671	0.4648	0.8335	0.2482	1.6871	0.8335	0.8353	0.0630	0.0387
0.1112	14.0	9338	0.4777	0.8317	0.2509	1.6393	0.8317	0.8312	0.0614	0.0418
0.1002	15.0	10005	0.4684	0.8333	0.2484	1.6054	0.8333	0.8335	0.0657	0.0392
0.0944	16.0	10672	0.4693	0.8365	0.2480	1.6381	0.8365	0.8366	0.0658	0.0383
0.0934	17.0	11339	0.4534	0.8323	0.2465	1.6420	0.8323	0.8343	0.0561	0.0373
0.0835	18.0	12006	0.4512	0.8357	0.2435	1.6301	0.8357	0.8367	0.0575	0.0372
0.08	19.0	12673	0.4345	0.838	0.2394	1.6382	0.838	0.8398	0.0562	0.0366
0.0819	20.0	13340	0.4356	0.838	0.2374	1.5973	0.838	0.8384	0.0588	0.0364
0.0709	21.0	14007	0.4484	0.8415	0.2368	1.6231	0.8415	0.8411	0.0595	0.0368
0.0691	22.0	14674	0.4194	0.8495	0.2287	1.5968	0.8495	0.8505	0.0531	0.0335
0.068	23.0	15341	0.4308	0.8413	0.2346	1.5599	0.8413	0.8410	0.0542	0.0360
0.0641	24.0	16008	0.4209	0.8405	0.2336	1.5539	0.8405	0.8422	0.0590	0.0339
0.0617	25.0	16675	0.4181	0.841	0.2352	1.5735	0.841	0.8435	0.0568	0.0356
0.0633	26.0	17342	0.4193	0.8508	0.2286	1.5299	0.8508	0.8510	0.0650	0.0348
0.0569	27.0	18009	0.4065	0.8468	0.2278	1.5267	0.8468	0.8479	0.0546	0.0332
0.0571	28.0	18676	0.4109	0.8498	0.2255	1.5147	0.8498	0.8499	0.0590	0.0331
0.0543	29.0	19343	0.4026	0.8482	0.2250	1.5187	0.8482	0.8498	0.0623	0.0327
0.0543	30.0	20010	0.4124	0.847	0.2293	1.5125	0.847	0.8473	0.0605	0.0330
0.0536	31.0	20677	0.4022	0.851	0.2238	1.5100	0.851	0.8527	0.0594	0.0323
0.0522	32.0	21344	0.4120	0.8475	0.2290	1.5044	0.8475	0.8483	0.0633	0.0327
0.0493	33.0	22011	0.3990	0.8492	0.2258	1.5197	0.8492	0.8503	0.0589	0.0318
0.0512	34.0	22678	0.3983	0.85	0.2251	1.4644	0.85	0.8503	0.0597	0.0319
0.0517	35.0	23345	0.3969	0.8465	0.2257	1.4814	0.8465	0.8479	0.0630	0.0309
0.0477	36.0	24012	0.3939	0.8528	0.2237	1.4797	0.8528	0.8531	0.0604	0.0316
0.0482	37.0	24679	0.3934	0.852	0.2218	1.4595	0.852	0.8527	0.0613	0.0316
0.0481	38.0	25346	0.3930	0.8532	0.2217	1.4561	0.8532	0.8544	0.0593	0.0306
0.0477	39.0	26013	0.3875	0.8512	0.2202	1.4610	0.8512	0.8523	0.0609	0.0310
0.048	40.0	26680	0.3900	0.8538	0.2202	1.4541	0.8537	0.8546	0.0629	0.0307
0.0448	41.0	27347	0.3901	0.8525	0.2221	1.4519	0.8525	0.8532	0.0621	0.0308
0.0454	42.0	28014	0.3858	0.851	0.2186	1.4554	0.851	0.8519	0.0633	0.0298
0.0464	43.0	28681	0.3861	0.8528	0.2197	1.4516	0.8528	0.8535	0.0618	0.0307
0.0444	44.0	29348	0.3824	0.8548	0.2176	1.4288	0.8547	0.8557	0.0607	0.0299
0.0461	45.0	30015	0.3833	0.8555	0.2181	1.4330	0.8555	0.8566	0.0606	0.0302
0.0442	46.0	30682	0.3830	0.8552	0.2174	1.4358	0.8552	0.8560	0.0604	0.0302
0.0456	47.0	31349	0.3797	0.8552	0.2173	1.4264	0.8552	0.8560	0.0596	0.0297
0.0447	48.0	32016	0.3811	0.8558	0.2176	1.4273	0.8558	0.8566	0.0595	0.0300
0.0439	49.0	32683	0.3814	0.856	0.2176	1.4252	0.856	0.8568	0.0600	0.0300
0.0437	50.0	33350	0.3813	0.8558	0.2176	1.4251	0.8558	0.8566	0.0597	0.0299

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/cdip-small_rvl_cdip-NK1000_kd_test

cdip-small_rvl_cdip-NK1000_kd_test

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions