resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_CEKD_t1.0_a1.0

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: 1.7844
Accuracy: 0.742
Brier Loss: 0.4405
Nll: 2.8680
F1 Micro: 0.7420
F1 Macro: 0.7411
Ece: 0.1946
Aurc: 0.1002

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	2.7345	0.153	0.9327	8.3371	0.153	0.1246	0.0866	0.7933
2.6983	2.0	500	2.4500	0.4213	0.8816	4.7062	0.4213	0.3924	0.3073	0.4444
2.6983	3.0	750	1.7959	0.5012	0.7003	3.3576	0.5012	0.4758	0.1869	0.3051
1.7341	4.0	1000	1.3637	0.5985	0.5511	2.8818	0.5985	0.5868	0.1005	0.1935
1.7341	5.0	1250	1.1978	0.6498	0.4862	2.7546	0.6498	0.6471	0.0826	0.1500
1.0818	6.0	1500	1.0812	0.6853	0.4364	2.6325	0.6853	0.6845	0.0522	0.1217
1.0818	7.0	1750	1.0276	0.7013	0.4149	2.5542	0.7013	0.7003	0.0397	0.1108
0.7498	8.0	2000	0.9724	0.7133	0.3944	2.4773	0.7133	0.7129	0.0505	0.1040
0.7498	9.0	2250	0.9777	0.7248	0.3924	2.4916	0.7248	0.7242	0.0628	0.0992
0.5034	10.0	2500	1.0027	0.724	0.3976	2.4974	0.724	0.7250	0.0751	0.1032
0.5034	11.0	2750	0.9979	0.729	0.3913	2.5344	0.729	0.7295	0.0805	0.0988
0.3237	12.0	3000	1.0553	0.7192	0.4075	2.6242	0.7192	0.7193	0.0963	0.1072
0.3237	13.0	3250	1.1162	0.7175	0.4139	2.6543	0.7175	0.7185	0.1295	0.1093
0.2023	14.0	3500	1.1259	0.725	0.4140	2.6758	0.7250	0.7246	0.1237	0.1055
0.2023	15.0	3750	1.2728	0.7115	0.4381	2.8308	0.7115	0.7147	0.1464	0.1168
0.1264	16.0	4000	1.2664	0.7222	0.4296	2.8434	0.7223	0.7236	0.1523	0.1107
0.1264	17.0	4250	1.2620	0.724	0.4252	2.7990	0.724	0.7252	0.1563	0.1066
0.0802	18.0	4500	1.3362	0.727	0.4293	2.8642	0.7270	0.7267	0.1653	0.1090
0.0802	19.0	4750	1.3608	0.7302	0.4288	2.7893	0.7302	0.7318	0.1637	0.1059
0.0553	20.0	5000	1.3757	0.7308	0.4303	2.8861	0.7308	0.7300	0.1670	0.1073
0.0553	21.0	5250	1.4947	0.7295	0.4420	2.8306	0.7295	0.7300	0.1770	0.1128
0.0329	22.0	5500	1.5338	0.7265	0.4416	2.8729	0.7265	0.7273	0.1808	0.1097
0.0329	23.0	5750	1.5127	0.7355	0.4362	2.8574	0.7355	0.7366	0.1774	0.1045
0.0258	24.0	6000	1.5189	0.7352	0.4360	2.8435	0.7353	0.7344	0.1784	0.1030
0.0258	25.0	6250	1.5802	0.7362	0.4404	2.8399	0.7362	0.7362	0.1847	0.1013
0.0193	26.0	6500	1.5869	0.737	0.4378	2.8237	0.737	0.7362	0.1846	0.1022
0.0193	27.0	6750	1.6160	0.7365	0.4373	2.7928	0.7365	0.7360	0.1864	0.1049
0.014	28.0	7000	1.6775	0.7372	0.4426	2.9236	0.7372	0.7373	0.1909	0.1039
0.014	29.0	7250	1.6391	0.736	0.4370	2.8717	0.736	0.7358	0.1905	0.0999
0.0132	30.0	7500	1.6804	0.7355	0.4434	2.8397	0.7355	0.7360	0.1903	0.1067
0.0132	31.0	7750	1.6809	0.738	0.4386	2.8853	0.738	0.7387	0.1920	0.1015
0.0121	32.0	8000	1.6953	0.734	0.4443	2.8451	0.734	0.7342	0.1961	0.1013
0.0121	33.0	8250	1.7184	0.7425	0.4344	2.8180	0.7425	0.7423	0.1910	0.1014
0.0098	34.0	8500	1.7151	0.735	0.4445	2.8532	0.735	0.7337	0.1952	0.1000
0.0098	35.0	8750	1.7781	0.7338	0.4484	2.8133	0.7338	0.7351	0.1999	0.1052
0.0086	36.0	9000	1.7540	0.7372	0.4443	2.8388	0.7372	0.7388	0.1954	0.1039
0.0086	37.0	9250	1.7744	0.738	0.4474	2.8600	0.738	0.7390	0.1953	0.1057
0.0079	38.0	9500	1.7446	0.7368	0.4417	2.8485	0.7367	0.7374	0.1972	0.1016
0.0079	39.0	9750	1.7700	0.739	0.4398	2.8826	0.739	0.7395	0.1970	0.1023
0.0076	40.0	10000	1.7896	0.7368	0.4442	2.8449	0.7367	0.7376	0.1988	0.1033
0.0076	41.0	10250	1.7435	0.7402	0.4387	2.8390	0.7402	0.7405	0.1926	0.1031
0.0074	42.0	10500	1.7837	0.7338	0.4470	2.8191	0.7338	0.7339	0.2018	0.1035
0.0074	43.0	10750	1.8015	0.7392	0.4427	2.8093	0.7392	0.7401	0.1981	0.1017
0.0061	44.0	11000	1.8155	0.739	0.4449	2.8333	0.739	0.7406	0.1983	0.1022
0.0061	45.0	11250	1.7958	0.7392	0.4426	2.8161	0.7392	0.7385	0.1963	0.1039
0.0059	46.0	11500	1.8089	0.7422	0.4411	2.8174	0.7422	0.7422	0.1955	0.1011
0.0059	47.0	11750	1.8125	0.743	0.4386	2.8184	0.743	0.7435	0.1939	0.1012
0.0053	48.0	12000	1.8004	0.7372	0.4432	2.8413	0.7372	0.7371	0.1995	0.1023
0.0053	49.0	12250	1.8075	0.7405	0.4392	2.8569	0.7405	0.7397	0.1962	0.1015
0.0055	50.0	12500	1.7844	0.742	0.4405	2.8680	0.7420	0.7411	0.1946	0.1002

Framework versions

Transformers 4.33.3
Pytorch 2.2.0.dev20231002
Datasets 2.7.1
Tokenizers 0.13.3

bdpc/resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_CEKD_t1.0_a1.0

resnet101_rvl-cdip-cnn_rvl_cdip-NK1000_kd_CEKD_t1.0_a1.0

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Seed this model