resnet101-base_tobacco-cnn_tobacco3482_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: 0.0899
• Accuracy: 0.395
• Brier Loss: 0.6867
• Nll: 4.7352
• F1 Micro: 0.395
• F1 Macro: 0.2347
• Ece: 0.2366
• Aurc: 0.3626

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	1.1202	0.17	0.8964	8.4790	0.17	0.1089	0.2136	0.8244
No log	2.0	26	1.0772	0.165	0.8950	8.2397	0.165	0.0929	0.2120	0.8534
No log	3.0	39	0.9427	0.2	0.8847	7.1036	0.2000	0.0796	0.2384	0.7748
No log	4.0	52	0.7947	0.21	0.8720	6.5481	0.2100	0.0649	0.2432	0.7270
No log	5.0	65	0.5378	0.205	0.8432	6.3064	0.205	0.0544	0.2367	0.6763
No log	6.0	78	0.4557	0.18	0.8402	6.3878	0.18	0.0308	0.2384	0.7467
No log	7.0	91	0.4326	0.18	0.8383	6.3386	0.18	0.0308	0.2385	0.7234
No log	8.0	104	0.2832	0.265	0.8085	6.3561	0.265	0.1012	0.2570	0.6272
No log	9.0	117	0.2672	0.255	0.8124	6.2296	0.255	0.0981	0.2569	0.6567
No log	10.0	130	0.2452	0.29	0.7953	6.3199	0.29	0.1153	0.2717	0.5884
No log	11.0	143	0.2155	0.31	0.7764	6.3618	0.31	0.1231	0.2728	0.4803
No log	12.0	156	0.1315	0.31	0.7371	6.2610	0.31	0.1231	0.2343	0.4419
No log	13.0	169	0.1803	0.3	0.7665	6.1189	0.3	0.1187	0.2587	0.4579
No log	14.0	182	0.1426	0.31	0.7386	6.1115	0.31	0.1236	0.2502	0.4341
No log	15.0	195	0.1431	0.31	0.7334	5.9353	0.31	0.1274	0.2624	0.4233
No log	16.0	208	0.1540	0.32	0.7318	5.7102	0.32	0.1432	0.2493	0.4322
No log	17.0	221	0.2603	0.305	0.7784	5.6776	0.305	0.1361	0.2751	0.5118
No log	18.0	234	0.1000	0.35	0.7074	5.4636	0.35	0.1574	0.2420	0.4027
No log	19.0	247	0.1014	0.33	0.7131	5.5297	0.33	0.1413	0.2439	0.4245
No log	20.0	260	0.2862	0.265	0.8013	5.5041	0.265	0.1126	0.2762	0.6324
No log	21.0	273	0.1224	0.34	0.7183	5.2027	0.34	0.1544	0.2673	0.4222
No log	22.0	286	0.1406	0.345	0.7173	5.1426	0.345	0.1612	0.2710	0.4019
No log	23.0	299	0.1509	0.34	0.7270	5.0281	0.34	0.1565	0.2641	0.4178
No log	24.0	312	0.0994	0.37	0.6996	5.1278	0.37	0.1771	0.2390	0.3930
No log	25.0	325	0.1965	0.35	0.7474	5.0356	0.35	0.1707	0.2774	0.4503
No log	26.0	338	0.1104	0.37	0.7085	5.0275	0.37	0.1984	0.2663	0.3927
No log	27.0	351	0.1674	0.34	0.7299	4.9200	0.34	0.1739	0.2787	0.4257
No log	28.0	364	0.2424	0.335	0.7626	5.0286	0.335	0.1693	0.2905	0.5297
No log	29.0	377	0.1261	0.345	0.7185	5.0591	0.345	0.1730	0.2892	0.4142
No log	30.0	390	0.1574	0.365	0.7213	4.8809	0.3650	0.1951	0.2983	0.4062
No log	31.0	403	0.1227	0.365	0.7098	4.8152	0.3650	0.1996	0.2802	0.3992
No log	32.0	416	0.1114	0.355	0.7010	4.8224	0.3550	0.1915	0.2657	0.3958
No log	33.0	429	0.1027	0.39	0.6934	4.7755	0.39	0.2245	0.2653	0.3695
No log	34.0	442	0.0959	0.385	0.6875	4.8715	0.3850	0.2299	0.2591	0.3699
No log	35.0	455	0.0905	0.395	0.6897	4.8649	0.395	0.2367	0.2519	0.3627
No log	36.0	468	0.0879	0.365	0.6911	4.8472	0.3650	0.2132	0.2437	0.3910
No log	37.0	481	0.0867	0.39	0.6881	4.7379	0.39	0.2335	0.2576	0.3680
No log	38.0	494	0.0934	0.4	0.6916	4.6797	0.4000	0.2490	0.2578	0.3628
0.2032	39.0	507	0.0928	0.38	0.6901	4.6734	0.38	0.2268	0.2432	0.3783
0.2032	40.0	520	0.0995	0.39	0.6875	4.8180	0.39	0.2323	0.2647	0.3730
0.2032	41.0	533	0.0944	0.37	0.6892	4.8193	0.37	0.2174	0.2536	0.3862
0.2032	42.0	546	0.0904	0.415	0.6885	4.5644	0.415	0.2556	0.2729	0.3573
0.2032	43.0	559	0.0951	0.39	0.6899	4.6549	0.39	0.2417	0.2525	0.3692
0.2032	44.0	572	0.0884	0.4	0.6860	4.6572	0.4000	0.2402	0.2587	0.3557
0.2032	45.0	585	0.0867	0.38	0.6874	4.6558	0.38	0.2278	0.2526	0.3738
0.2032	46.0	598	0.0861	0.405	0.6844	4.5777	0.405	0.2537	0.2548	0.3628
0.2032	47.0	611	0.0874	0.385	0.6853	4.4946	0.3850	0.2380	0.2570	0.3743
0.2032	48.0	624	0.0880	0.405	0.6857	4.5605	0.405	0.2500	0.2489	0.3555
0.2032	49.0	637	0.0884	0.4	0.6853	4.6057	0.4000	0.2481	0.2401	0.3616
0.2032	50.0	650	0.0899	0.395	0.6867	4.7352	0.395	0.2347	0.2366	0.3626

Framework versions

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