resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_simkd_rand

This model is a fine-tuned version of bdpc/resnet101_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4174
• Accuracy: 0.7665
• Brier Loss: 0.3263
• Nll: 2.0962
• F1 Micro: 0.7665
• F1 Macro: 0.7661
• Ece: 0.0504
• Aurc: 0.0700

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	0.8645	0.1192	0.9514	3.2233	0.1192	0.0652	0.1115	0.8122
1.2523	2.0	500	0.7139	0.1797	0.8939	3.1527	0.1798	0.1283	0.0795	0.6807
1.2523	3.0	750	0.6662	0.3145	0.8040	6.3258	0.3145	0.2485	0.0647	0.4987
0.6553	4.0	1000	0.6265	0.3738	0.7356	6.0830	0.3738	0.3459	0.0768	0.4070
0.6553	5.0	1250	0.5609	0.531	0.6047	4.7056	0.531	0.5234	0.0639	0.2463
0.5525	6.0	1500	0.5341	0.589	0.5450	3.9772	0.589	0.5948	0.0718	0.1912
0.5525	7.0	1750	0.4938	0.6468	0.4733	3.3676	0.6468	0.6486	0.0670	0.1408
0.4842	8.0	2000	0.4765	0.7	0.4288	2.8692	0.7	0.6960	0.0666	0.1181
0.4842	9.0	2250	0.5359	0.5938	0.5534	3.9887	0.5938	0.6011	0.1211	0.1809
0.4476	10.0	2500	0.4611	0.7037	0.4122	2.7429	0.7037	0.6991	0.0679	0.1097
0.4476	11.0	2750	0.4460	0.7225	0.3913	2.6158	0.7225	0.7240	0.0725	0.0967
0.4219	12.0	3000	0.4387	0.7388	0.3752	2.4639	0.7388	0.7356	0.0696	0.0892
0.4219	13.0	3250	0.4399	0.7378	0.3724	2.4683	0.7378	0.7381	0.0550	0.0898
0.4007	14.0	3500	0.4441	0.737	0.3738	2.4680	0.737	0.7334	0.0581	0.0906
0.4007	15.0	3750	0.4517	0.7248	0.3906	2.5901	0.7248	0.7302	0.0653	0.0961
0.3825	16.0	4000	0.4430	0.737	0.3727	2.5633	0.737	0.7350	0.0595	0.0884
0.3825	17.0	4250	0.4345	0.7482	0.3518	2.3938	0.7482	0.7473	0.0541	0.0784
0.3672	18.0	4500	0.4642	0.7385	0.3690	2.4016	0.7385	0.7367	0.0571	0.0891
0.3672	19.0	4750	0.4309	0.7432	0.3585	2.3331	0.7432	0.7464	0.0558	0.0824
0.3547	20.0	5000	0.4205	0.7602	0.3418	2.2097	0.7602	0.7617	0.0470	0.0744
0.3547	21.0	5250	0.4174	0.7602	0.3387	2.2020	0.7602	0.7594	0.0488	0.0748
0.3442	22.0	5500	0.4207	0.7515	0.3458	2.2370	0.7515	0.7543	0.0540	0.0777
0.3442	23.0	5750	0.4465	0.733	0.3783	2.5113	0.733	0.7295	0.0576	0.0919
0.3355	24.0	6000	0.4391	0.7425	0.3649	2.4598	0.7425	0.7459	0.0534	0.0830
0.3355	25.0	6250	0.4233	0.7598	0.3352	2.2321	0.7598	0.7609	0.0495	0.0729
0.3274	26.0	6500	0.4174	0.7665	0.3305	2.2062	0.7665	0.7673	0.0482	0.0699
0.3274	27.0	6750	0.4153	0.7598	0.3389	2.2158	0.7598	0.7583	0.0549	0.0740
0.3206	28.0	7000	0.4175	0.763	0.3323	2.1843	0.763	0.7610	0.0494	0.0721
0.3206	29.0	7250	0.4201	0.7522	0.3467	2.2627	0.7522	0.7495	0.0576	0.0783
0.3147	30.0	7500	0.4133	0.7625	0.3334	2.1459	0.7625	0.7631	0.0477	0.0733
0.3147	31.0	7750	0.4213	0.7558	0.3421	2.2877	0.7558	0.7535	0.0567	0.0758
0.3092	32.0	8000	0.4136	0.7668	0.3294	2.1791	0.7668	0.7662	0.0465	0.0702
0.3092	33.0	8250	0.4114	0.7638	0.3331	2.1993	0.7638	0.7613	0.0517	0.0722
0.3046	34.0	8500	0.4154	0.764	0.3294	2.1689	0.764	0.7639	0.0489	0.0714
0.3046	35.0	8750	0.4119	0.7638	0.3327	2.1482	0.7638	0.7628	0.0449	0.0725
0.3001	36.0	9000	0.4183	0.759	0.3348	2.1775	0.7590	0.7605	0.0513	0.0731
0.3001	37.0	9250	0.4097	0.7578	0.3344	2.2029	0.7577	0.7571	0.0525	0.0736
0.2964	38.0	9500	0.4126	0.7655	0.3292	2.1374	0.7655	0.7657	0.0481	0.0710
0.2964	39.0	9750	0.4235	0.7642	0.3287	2.1640	0.7642	0.7639	0.0543	0.0707
0.293	40.0	10000	0.4168	0.7678	0.3284	2.1264	0.7678	0.7681	0.0494	0.0702
0.293	41.0	10250	0.4118	0.7682	0.3270	2.1387	0.7682	0.7684	0.0462	0.0702
0.29	42.0	10500	0.4151	0.7618	0.3288	2.1464	0.7618	0.7609	0.0493	0.0718
0.29	43.0	10750	0.4172	0.7608	0.3283	2.1341	0.7608	0.7607	0.0538	0.0708
0.2876	44.0	11000	0.4159	0.7612	0.3278	2.1561	0.7612	0.7601	0.0514	0.0707
0.2876	45.0	11250	0.4173	0.761	0.3291	2.1825	0.761	0.7602	0.0493	0.0711
0.2855	46.0	11500	0.4137	0.761	0.3295	2.1514	0.761	0.7598	0.0507	0.0709
0.2855	47.0	11750	0.4143	0.764	0.3278	2.1414	0.764	0.7630	0.0483	0.0705
0.2841	48.0	12000	0.4162	0.7668	0.3262	2.1191	0.7668	0.7666	0.0451	0.0699
0.2841	49.0	12250	0.4190	0.765	0.3271	2.1267	0.765	0.7647	0.0496	0.0701
0.283	50.0	12500	0.4174	0.7665	0.3263	2.0962	0.7665	0.7661	0.0504	0.0700

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2