resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t5.0_a0.5

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: 0.6454
• Accuracy: 0.685
• Brier Loss: 0.4931
• Nll: 2.5040
• F1 Micro: 0.685
• F1 Macro: 0.6171
• Ece: 0.2996
• Aurc: 0.1499

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	13	1.4475	0.17	0.8966	8.4781	0.17	0.1126	0.2169	0.8299
No log	2.0	26	1.4360	0.165	0.8955	8.4118	0.165	0.1097	0.2115	0.8359
No log	3.0	39	1.3776	0.16	0.8842	6.1685	0.16	0.0633	0.2066	0.7780
No log	4.0	52	1.3085	0.2	0.8701	6.0521	0.2000	0.0728	0.2332	0.7424
No log	5.0	65	1.2551	0.18	0.8597	6.1887	0.18	0.0491	0.2265	0.7890
No log	6.0	78	1.2118	0.2	0.8489	6.1706	0.2000	0.0631	0.2324	0.7179
No log	7.0	91	1.1759	0.19	0.8418	6.1310	0.19	0.0428	0.2384	0.7431
No log	8.0	104	1.1577	0.195	0.8339	5.7305	0.195	0.0535	0.2399	0.7126
No log	9.0	117	0.9905	0.34	0.7692	6.1092	0.34	0.1567	0.2772	0.4320
No log	10.0	130	0.9603	0.355	0.7541	5.7998	0.3550	0.1969	0.3021	0.4111
No log	11.0	143	1.0839	0.255	0.8087	5.1464	0.255	0.1242	0.2389	0.6769
No log	12.0	156	0.9374	0.39	0.7410	4.8415	0.39	0.2220	0.3037	0.4194
No log	13.0	169	0.9974	0.33	0.7720	4.9023	0.33	0.1732	0.2863	0.6049
No log	14.0	182	0.9393	0.435	0.7251	4.3102	0.435	0.2455	0.3276	0.3645
No log	15.0	195	0.9554	0.39	0.7416	4.0361	0.39	0.2535	0.2721	0.5075
No log	16.0	208	0.8012	0.465	0.6445	4.0129	0.465	0.2935	0.2551	0.2839
No log	17.0	221	0.8033	0.53	0.6418	3.4959	0.53	0.3816	0.3242	0.2458
No log	18.0	234	0.7740	0.57	0.6204	3.4062	0.57	0.4297	0.3139	0.2245
No log	19.0	247	0.7736	0.6	0.6124	3.3460	0.6	0.4408	0.3017	0.1919
No log	20.0	260	0.9105	0.555	0.6919	3.2115	0.555	0.4524	0.3604	0.3099
No log	21.0	273	0.7416	0.61	0.5948	3.1349	0.61	0.5093	0.3176	0.2233
No log	22.0	286	0.7318	0.655	0.5815	3.1259	0.655	0.5433	0.3478	0.1672
No log	23.0	299	0.7799	0.59	0.6079	3.0590	0.59	0.4963	0.3340	0.2455
No log	24.0	312	0.7886	0.665	0.6038	2.9965	0.665	0.5575	0.3773	0.1623
No log	25.0	325	0.7083	0.66	0.5602	3.0752	0.66	0.5582	0.3283	0.1772
No log	26.0	338	0.6882	0.63	0.5507	2.9022	0.63	0.5404	0.2963	0.1851
No log	27.0	351	0.6774	0.66	0.5348	2.7876	0.66	0.5674	0.3095	0.1662
No log	28.0	364	0.8111	0.675	0.6067	2.7578	0.675	0.5800	0.3905	0.1923
No log	29.0	377	0.6803	0.645	0.5338	2.8666	0.645	0.5486	0.3054	0.1646
No log	30.0	390	0.6835	0.685	0.5336	2.5944	0.685	0.5840	0.3119	0.1595
No log	31.0	403	0.6810	0.655	0.5309	2.7112	0.655	0.5625	0.2879	0.1786
No log	32.0	416	0.6848	0.685	0.5194	2.6456	0.685	0.5893	0.3314	0.1350
No log	33.0	429	0.6631	0.695	0.5063	2.6286	0.695	0.5980	0.3198	0.1314
No log	34.0	442	0.6639	0.69	0.5126	2.3890	0.69	0.5834	0.2990	0.1376
No log	35.0	455	0.6736	0.675	0.5172	2.3291	0.675	0.6014	0.3148	0.1646
No log	36.0	468	0.6648	0.68	0.5137	2.4549	0.68	0.6156	0.3316	0.1492
No log	37.0	481	0.6543	0.7	0.5006	2.4275	0.7	0.6130	0.3041	0.1342
No log	38.0	494	0.6514	0.675	0.5001	2.4064	0.675	0.5984	0.2963	0.1491
0.7462	39.0	507	0.6498	0.71	0.4988	2.5772	0.7100	0.6405	0.2980	0.1335
0.7462	40.0	520	0.6496	0.705	0.4964	2.5649	0.705	0.6386	0.3060	0.1380
0.7462	41.0	533	0.6562	0.68	0.5027	2.5816	0.68	0.6026	0.3100	0.1467
0.7462	42.0	546	0.6632	0.68	0.5089	2.4570	0.68	0.6112	0.2989	0.1500
0.7462	43.0	559	0.6437	0.7	0.4885	2.3648	0.7	0.6331	0.2741	0.1427
0.7462	44.0	572	0.6435	0.705	0.4894	2.4253	0.705	0.6370	0.3043	0.1390
0.7462	45.0	585	0.6457	0.695	0.4929	2.3611	0.695	0.6314	0.3021	0.1449
0.7462	46.0	598	0.6437	0.695	0.4912	2.3639	0.695	0.6370	0.2984	0.1436
0.7462	47.0	611	0.6466	0.685	0.4933	2.4859	0.685	0.6306	0.2936	0.1474
0.7462	48.0	624	0.6470	0.67	0.4950	2.3782	0.67	0.6070	0.3139	0.1547
0.7462	49.0	637	0.6477	0.675	0.4945	2.4509	0.675	0.6092	0.2852	0.1527
0.7462	50.0	650	0.6454	0.685	0.4931	2.5040	0.685	0.6171	0.2996	0.1499

Framework versions

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