resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_og_simkd

This model is a fine-tuned version of microsoft/resnet-50 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3748
• Accuracy: 0.8023
• Brier Loss: 0.2845
• Nll: 1.8818
• F1 Micro: 0.8023
• F1 Macro: 0.8020
• Ece: 0.0375
• Aurc: 0.0534

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: 64
• eval_batch_size: 64
• 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
No log	1.0	250	0.8880	0.1955	0.8872	5.3865	0.1955	0.1551	0.0582	0.7111
0.9199	2.0	500	0.6464	0.407	0.7284	5.2363	0.4070	0.3745	0.0770	0.4284
0.9199	3.0	750	0.5608	0.5945	0.5337	3.5976	0.5945	0.5912	0.0561	0.1950
0.563	4.0	1000	0.4962	0.6905	0.4235	2.6948	0.6905	0.6885	0.0474	0.1170
0.563	5.0	1250	0.4613	0.7177	0.3858	2.5472	0.7178	0.7181	0.0512	0.0964
0.4567	6.0	1500	0.4372	0.742	0.3584	2.3396	0.7420	0.7425	0.0527	0.0824
0.4567	7.0	1750	0.4271	0.7595	0.3406	2.2123	0.7595	0.7596	0.0459	0.0756
0.4103	8.0	2000	0.4129	0.7658	0.3308	2.1667	0.7658	0.7666	0.0439	0.0704
0.4103	9.0	2250	0.4070	0.7678	0.3296	2.1663	0.7678	0.7692	0.0485	0.0699
0.3836	10.0	2500	0.4017	0.7725	0.3209	2.1207	0.7725	0.7732	0.0426	0.0667
0.3836	11.0	2750	0.3984	0.7768	0.3153	2.0353	0.7768	0.7771	0.0454	0.0651
0.3645	12.0	3000	0.3961	0.7752	0.3124	2.0755	0.7752	0.7754	0.0428	0.0642
0.3645	13.0	3250	0.3961	0.786	0.3071	1.9949	0.786	0.7861	0.0407	0.0612
0.3497	14.0	3500	0.3899	0.7823	0.3053	1.9769	0.7823	0.7823	0.0435	0.0606
0.3497	15.0	3750	0.3873	0.7853	0.3021	1.9881	0.7853	0.7849	0.0479	0.0594
0.3378	16.0	4000	0.3861	0.7833	0.3026	1.9263	0.7833	0.7834	0.0431	0.0593
0.3378	17.0	4250	0.3853	0.7913	0.2970	1.9108	0.7913	0.7917	0.0390	0.0571
0.3271	18.0	4500	0.3840	0.7903	0.2978	1.9643	0.7903	0.7902	0.0377	0.0576
0.3271	19.0	4750	0.3828	0.7915	0.2967	1.9332	0.7915	0.7914	0.0393	0.0572
0.3186	20.0	5000	0.3806	0.7913	0.2938	1.9410	0.7913	0.7909	0.0410	0.0563
0.3186	21.0	5250	0.3815	0.7953	0.2921	1.9285	0.7953	0.7949	0.0387	0.0566
0.3111	22.0	5500	0.3838	0.7895	0.2949	1.9126	0.7895	0.7894	0.0382	0.0570
0.3111	23.0	5750	0.3799	0.7955	0.2902	1.9332	0.7955	0.7955	0.0373	0.0558
0.305	24.0	6000	0.3796	0.7947	0.2912	1.8615	0.7947	0.7940	0.0418	0.0561
0.305	25.0	6250	0.3805	0.7947	0.2912	1.8999	0.7947	0.7940	0.0413	0.0558
0.2993	26.0	6500	0.3842	0.7925	0.2913	1.9451	0.7925	0.7927	0.0339	0.0559
0.2993	27.0	6750	0.3784	0.794	0.2908	1.9151	0.7940	0.7942	0.0389	0.0553
0.2943	28.0	7000	0.3779	0.7957	0.2895	1.8758	0.7957	0.7957	0.0392	0.0549
0.2943	29.0	7250	0.3776	0.7955	0.2892	1.8785	0.7955	0.7947	0.0445	0.0549
0.2905	30.0	7500	0.3775	0.7973	0.2879	1.8786	0.7973	0.7972	0.0379	0.0550
0.2905	31.0	7750	0.3773	0.7945	0.2903	1.9039	0.7945	0.7942	0.0405	0.0551
0.2863	32.0	8000	0.3764	0.7963	0.2880	1.8569	0.7963	0.7962	0.0375	0.0549
0.2863	33.0	8250	0.3775	0.7925	0.2884	1.9070	0.7925	0.7917	0.0411	0.0544
0.2831	34.0	8500	0.3762	0.7935	0.2873	1.8608	0.7935	0.7933	0.0389	0.0547
0.2831	35.0	8750	0.3765	0.7973	0.2868	1.9316	0.7973	0.7970	0.0385	0.0540
0.28	36.0	9000	0.3750	0.7967	0.2857	1.8871	0.7967	0.7965	0.0375	0.0540
0.28	37.0	9250	0.3761	0.793	0.2874	1.8977	0.793	0.7926	0.0405	0.0543
0.2775	38.0	9500	0.3760	0.7983	0.2861	1.8613	0.7983	0.7987	0.0422	0.0540
0.2775	39.0	9750	0.3761	0.7955	0.2870	1.8744	0.7955	0.7957	0.0412	0.0545
0.2755	40.0	10000	0.3753	0.8007	0.2852	1.8640	0.8007	0.8006	0.0345	0.0532
0.2755	41.0	10250	0.3753	0.8023	0.2857	1.8637	0.8023	0.8025	0.0363	0.0535
0.2735	42.0	10500	0.3751	0.7995	0.2851	1.9134	0.7995	0.7994	0.0403	0.0531
0.2735	43.0	10750	0.3753	0.8	0.2857	1.8832	0.8000	0.7996	0.0406	0.0538
0.2717	44.0	11000	0.3746	0.7985	0.2851	1.8545	0.7985	0.7982	0.0432	0.0532
0.2717	45.0	11250	0.3747	0.7985	0.2847	1.8730	0.7985	0.7984	0.0400	0.0534
0.2701	46.0	11500	0.3744	0.801	0.2843	1.8783	0.801	0.8007	0.0411	0.0532
0.2701	47.0	11750	0.3744	0.798	0.2852	1.8843	0.798	0.7975	0.0420	0.0535
0.2694	48.0	12000	0.3753	0.7993	0.2857	1.8875	0.7993	0.7988	0.0405	0.0532
0.2694	49.0	12250	0.3758	0.7965	0.2868	1.8927	0.7965	0.7964	0.0415	0.0539
0.2684	50.0	12500	0.3748	0.8023	0.2845	1.8818	0.8023	0.8020	0.0375	0.0534

Framework versions

• Transformers 4.33.3
• Pytorch 2.2.0.dev20231002
• Datasets 2.7.1
• Tokenizers 0.13.3