resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_MSE

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: 0.7429
• Accuracy: 0.7853
• Brier Loss: 0.3044
• Nll: 2.0364
• F1 Micro: 0.7853
• F1 Macro: 0.7862
• Ece: 0.0430
• Aurc: 0.0599

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	250	9.5443	0.0765	0.9365	3.7373	0.0765	0.0522	0.0360	0.9336
9.4095	2.0	500	7.4542	0.0757	0.9312	2.8468	0.0757	0.0316	0.0425	0.8840
9.4095	3.0	750	5.8933	0.0975	0.9356	3.2058	0.0975	0.0408	0.0798	0.8593
5.9994	4.0	1000	4.3665	0.2125	0.8700	5.3759	0.2125	0.1290	0.0743	0.7029
5.9994	5.0	1250	3.0367	0.4415	0.6924	4.9073	0.4415	0.4283	0.0806	0.3570
3.2184	6.0	1500	2.1589	0.579	0.5587	3.7412	0.579	0.5771	0.0572	0.2172
3.2184	7.0	1750	1.5582	0.652	0.4673	3.0701	0.652	0.6456	0.0517	0.1478
1.6737	8.0	2000	1.3502	0.6893	0.4266	2.8575	0.6893	0.6860	0.0544	0.1175
1.6737	9.0	2250	1.1389	0.7188	0.3914	2.5937	0.7188	0.7195	0.0544	0.1006
1.0789	10.0	2500	1.0563	0.7302	0.3742	2.5043	0.7302	0.7305	0.0618	0.0912
1.0789	11.0	2750	1.0035	0.7428	0.3604	2.4375	0.7428	0.7441	0.0587	0.0823
0.7934	12.0	3000	0.9169	0.7548	0.3472	2.2921	0.7548	0.7555	0.0547	0.0762
0.7934	13.0	3250	0.8628	0.7598	0.3386	2.2849	0.7598	0.7600	0.0550	0.0739
0.6268	14.0	3500	0.8773	0.7675	0.3362	2.2170	0.7675	0.7692	0.0490	0.0718
0.6268	15.0	3750	0.8263	0.7682	0.3306	2.1617	0.7682	0.7702	0.0534	0.0704
0.5269	16.0	4000	0.8422	0.7708	0.3289	2.1907	0.7707	0.7717	0.0524	0.0687
0.5269	17.0	4250	0.8100	0.7745	0.3241	2.1664	0.7745	0.7761	0.0509	0.0667
0.4516	18.0	4500	0.8013	0.7778	0.3215	2.1216	0.7778	0.7790	0.0473	0.0669
0.4516	19.0	4750	0.7911	0.7802	0.3183	2.1224	0.7802	0.7812	0.0476	0.0648
0.4039	20.0	5000	0.7900	0.7775	0.3197	2.0969	0.7775	0.7797	0.0473	0.0647
0.4039	21.0	5250	0.7919	0.7792	0.3191	2.1445	0.7792	0.7810	0.0531	0.0652
0.3563	22.0	5500	0.7960	0.7802	0.3166	2.0849	0.7802	0.7818	0.0478	0.0649
0.3563	23.0	5750	0.7615	0.7825	0.3128	2.0834	0.7825	0.7833	0.0478	0.0638
0.3251	24.0	6000	0.7840	0.7792	0.3151	2.0841	0.7792	0.7800	0.0513	0.0648
0.3251	25.0	6250	0.7837	0.7792	0.3159	2.0889	0.7792	0.7808	0.0485	0.0643
0.2949	26.0	6500	0.7827	0.7802	0.3158	2.0416	0.7802	0.7819	0.0548	0.0648
0.2949	27.0	6750	0.7650	0.78	0.3130	2.0411	0.78	0.7807	0.0506	0.0629
0.2669	28.0	7000	0.7787	0.7802	0.3133	2.0843	0.7802	0.7810	0.0454	0.0627
0.2669	29.0	7250	0.7892	0.782	0.3163	2.0953	0.782	0.7826	0.0508	0.0635
0.2512	30.0	7500	0.7775	0.7825	0.3126	2.0904	0.7825	0.7837	0.0451	0.0633
0.2512	31.0	7750	0.7601	0.7817	0.3124	2.0251	0.7817	0.7827	0.0485	0.0627
0.231	32.0	8000	0.7669	0.7833	0.3120	2.0685	0.7833	0.7842	0.0472	0.0629
0.231	33.0	8250	0.7652	0.7847	0.3116	2.0661	0.7847	0.7858	0.0519	0.0625
0.2172	34.0	8500	0.7637	0.7837	0.3107	2.0264	0.7837	0.7852	0.0487	0.0628
0.2172	35.0	8750	0.7691	0.783	0.3120	2.0535	0.7830	0.7844	0.0438	0.0634
0.2032	36.0	9000	0.7647	0.7845	0.3093	2.0480	0.7845	0.7852	0.0471	0.0620
0.2032	37.0	9250	0.7727	0.782	0.3122	2.0610	0.782	0.7830	0.0493	0.0628
0.1925	38.0	9500	0.7563	0.7843	0.3085	2.0267	0.7843	0.7849	0.0459	0.0608
0.1925	39.0	9750	0.7597	0.7835	0.3087	2.0062	0.7835	0.7845	0.0485	0.0614
0.1823	40.0	10000	0.7611	0.7833	0.3107	2.0007	0.7833	0.7853	0.0479	0.0625
0.1823	41.0	10250	0.7608	0.7843	0.3076	2.0335	0.7843	0.7854	0.0486	0.0602
0.17	42.0	10500	0.7535	0.7833	0.3096	2.0121	0.7833	0.7844	0.0505	0.0613
0.17	43.0	10750	0.7524	0.7845	0.3066	2.0425	0.7845	0.7856	0.0476	0.0605
0.1639	44.0	11000	0.7608	0.7808	0.3108	2.0739	0.7808	0.7816	0.0503	0.0618
0.1639	45.0	11250	0.7560	0.786	0.3063	1.9876	0.786	0.7868	0.0496	0.0607
0.1575	46.0	11500	0.7494	0.784	0.3063	2.0311	0.7840	0.7846	0.0416	0.0601
0.1575	47.0	11750	0.7515	0.7857	0.3069	2.0539	0.7857	0.7866	0.0456	0.0609
0.1493	48.0	12000	0.7511	0.7843	0.3086	2.0325	0.7843	0.7852	0.0552	0.0612
0.1493	49.0	12250	0.7495	0.787	0.3067	2.0231	0.787	0.7880	0.0475	0.0605
0.1425	50.0	12500	0.7538	0.7867	0.3052	2.0267	0.7868	0.7870	0.0507	0.0603
0.1425	51.0	12750	0.7529	0.7847	0.3081	2.0592	0.7847	0.7859	0.0467	0.0604
0.1356	52.0	13000	0.7527	0.7808	0.3071	2.0349	0.7808	0.7818	0.0473	0.0607
0.1356	53.0	13250	0.7451	0.7865	0.3049	2.0368	0.7865	0.7879	0.0484	0.0595
0.1325	54.0	13500	0.7481	0.7857	0.3056	2.0223	0.7857	0.7869	0.0468	0.0603
0.1325	55.0	13750	0.7470	0.7835	0.3057	2.0306	0.7835	0.7844	0.0492	0.0601
0.1264	56.0	14000	0.7471	0.7873	0.3053	2.0336	0.7873	0.7880	0.0519	0.0601
0.1264	57.0	14250	0.7429	0.7895	0.3032	2.0149	0.7895	0.7903	0.0468	0.0595
0.1208	58.0	14500	0.7399	0.7885	0.3035	2.0147	0.7885	0.7895	0.0433	0.0596
0.1208	59.0	14750	0.7518	0.786	0.3076	2.0481	0.786	0.7873	0.0403	0.0607
0.119	60.0	15000	0.7483	0.7903	0.3058	2.0138	0.7903	0.7914	0.0471	0.0601
0.119	61.0	15250	0.7463	0.7845	0.3043	2.0617	0.7845	0.7855	0.0458	0.0599
0.1128	62.0	15500	0.7478	0.7875	0.3056	2.0187	0.7875	0.7888	0.0452	0.0604
0.1128	63.0	15750	0.7510	0.784	0.3061	2.0204	0.7840	0.7850	0.0495	0.0605
0.1109	64.0	16000	0.7424	0.786	0.3053	2.0167	0.786	0.7871	0.0449	0.0603
0.1109	65.0	16250	0.7473	0.7885	0.3054	2.0200	0.7885	0.7893	0.0471	0.0600
0.1078	66.0	16500	0.7467	0.7873	0.3054	2.0224	0.7873	0.7883	0.0482	0.0599
0.1078	67.0	16750	0.7445	0.7893	0.3039	2.0082	0.7893	0.7895	0.0456	0.0593
0.1051	68.0	17000	0.7490	0.7873	0.3063	2.0152	0.7873	0.7883	0.0505	0.0602
0.1051	69.0	17250	0.7490	0.785	0.3061	2.0103	0.785	0.7861	0.0465	0.0602
0.1009	70.0	17500	0.7445	0.7875	0.3049	2.0308	0.7875	0.7884	0.0483	0.0598
0.1009	71.0	17750	0.7490	0.7863	0.3068	2.0260	0.7863	0.7875	0.0495	0.0604
0.0984	72.0	18000	0.7465	0.7893	0.3059	2.0161	0.7893	0.7906	0.0427	0.0601
0.0984	73.0	18250	0.7451	0.7873	0.3058	2.0204	0.7873	0.7882	0.0511	0.0605
0.0966	74.0	18500	0.7445	0.7875	0.3042	2.0227	0.7875	0.7886	0.0495	0.0599
0.0966	75.0	18750	0.7443	0.7863	0.3040	2.0138	0.7863	0.7872	0.0442	0.0598
0.0947	76.0	19000	0.7448	0.7865	0.3054	2.0234	0.7865	0.7873	0.0457	0.0598
0.0947	77.0	19250	0.7448	0.7865	0.3041	2.0110	0.7865	0.7875	0.0508	0.0596
0.0931	78.0	19500	0.7460	0.7883	0.3040	2.0125	0.7883	0.7895	0.0467	0.0595
0.0931	79.0	19750	0.7456	0.7883	0.3038	2.0302	0.7883	0.7894	0.0455	0.0596
0.0899	80.0	20000	0.7469	0.788	0.3040	2.0188	0.788	0.7892	0.0487	0.0597
0.0899	81.0	20250	0.7421	0.788	0.3041	2.0359	0.788	0.7888	0.0427	0.0595
0.0882	82.0	20500	0.7444	0.7865	0.3051	2.0219	0.7865	0.7875	0.0479	0.0600
0.0882	83.0	20750	0.7439	0.788	0.3039	2.0197	0.788	0.7894	0.0439	0.0597
0.0871	84.0	21000	0.7421	0.7865	0.3040	1.9910	0.7865	0.7876	0.0445	0.0598
0.0871	85.0	21250	0.7429	0.7887	0.3043	2.0253	0.7887	0.7898	0.0426	0.0597
0.0869	86.0	21500	0.7442	0.7873	0.3041	2.0156	0.7873	0.7885	0.0488	0.0596
0.0869	87.0	21750	0.7439	0.7857	0.3051	2.0099	0.7857	0.7867	0.0465	0.0599
0.084	88.0	22000	0.7434	0.786	0.3040	1.9926	0.786	0.7869	0.0469	0.0598
0.084	89.0	22250	0.7431	0.7873	0.3048	2.0028	0.7873	0.7880	0.0442	0.0599
0.0821	90.0	22500	0.7447	0.7867	0.3040	2.0349	0.7868	0.7876	0.0477	0.0596
0.0821	91.0	22750	0.7443	0.7877	0.3051	2.0356	0.7877	0.7887	0.0486	0.0601
0.0813	92.0	23000	0.7500	0.7873	0.3053	2.0465	0.7873	0.7880	0.0484	0.0601
0.0813	93.0	23250	0.7449	0.788	0.3037	1.9966	0.788	0.7890	0.0441	0.0594
0.0811	94.0	23500	0.7466	0.7897	0.3048	2.0297	0.7897	0.7907	0.0429	0.0600
0.0811	95.0	23750	0.7482	0.7875	0.3058	2.0319	0.7875	0.7885	0.0464	0.0601
0.0808	96.0	24000	0.7473	0.7863	0.3055	2.0219	0.7863	0.7874	0.0477	0.0603
0.0808	97.0	24250	0.7451	0.7855	0.3044	2.0356	0.7855	0.7865	0.0481	0.0594
0.08	98.0	24500	0.7442	0.7857	0.3042	2.0213	0.7857	0.7868	0.0475	0.0595
0.08	99.0	24750	0.7462	0.7863	0.3053	2.0354	0.7863	0.7874	0.0425	0.0599
0.079	100.0	25000	0.7429	0.7853	0.3044	2.0364	0.7853	0.7862	0.0430	0.0599

Framework versions

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