resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t2.5_a0.9

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: 0.8672
• Accuracy: 0.71
• Brier Loss: 0.4047
• Nll: 2.1924
• F1 Micro: 0.7100
• F1 Macro: 0.6463
• Ece: 0.2420
• Aurc: 0.1050

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	2.1239	0.16	0.8967	8.4233	0.16	0.1062	0.2101	0.8304
No log	2.0	26	2.1201	0.14	0.8961	8.2220	0.14	0.0876	0.1970	0.8491
No log	3.0	39	2.0724	0.215	0.8865	6.2039	0.2150	0.1169	0.2432	0.7837
No log	4.0	52	2.0291	0.185	0.8773	5.6169	0.185	0.0792	0.2329	0.7651
No log	5.0	65	1.9592	0.215	0.8614	6.0237	0.2150	0.0835	0.2493	0.7373
No log	6.0	78	1.9039	0.205	0.8483	5.9575	0.205	0.0619	0.2493	0.7526
No log	7.0	91	1.8651	0.26	0.8381	5.6215	0.26	0.1490	0.2663	0.6747
No log	8.0	104	1.8342	0.225	0.8311	5.7631	0.225	0.1071	0.2425	0.6919
No log	9.0	117	1.8057	0.31	0.8218	5.2969	0.31	0.2118	0.2795	0.6489
No log	10.0	130	1.5737	0.46	0.7277	5.1748	0.46	0.2853	0.3279	0.2977
No log	11.0	143	1.5629	0.415	0.7331	4.8259	0.415	0.2846	0.2924	0.3880
No log	12.0	156	1.5283	0.45	0.7135	4.0012	0.45	0.3122	0.3298	0.3197
No log	13.0	169	1.4200	0.51	0.6674	3.9849	0.51	0.3400	0.3259	0.2549
No log	14.0	182	1.4334	0.535	0.6710	3.7006	0.535	0.3840	0.3291	0.2584
No log	15.0	195	1.4306	0.45	0.6854	3.8260	0.45	0.3120	0.3055	0.4297
No log	16.0	208	1.3175	0.585	0.6174	3.3484	0.585	0.4401	0.3406	0.1916
No log	17.0	221	1.2680	0.57	0.5998	3.1408	0.57	0.4356	0.2903	0.2136
No log	18.0	234	1.2605	0.58	0.6020	3.2085	0.58	0.4711	0.2915	0.2355
No log	19.0	247	1.2292	0.61	0.5791	3.0633	0.61	0.5021	0.2929	0.2082
No log	20.0	260	1.3872	0.54	0.6604	3.2778	0.54	0.4604	0.3284	0.3506
No log	21.0	273	1.1646	0.625	0.5520	2.8539	0.625	0.5193	0.2828	0.1885
No log	22.0	286	1.1565	0.655	0.5438	2.6915	0.655	0.5437	0.3430	0.1549
No log	23.0	299	1.1041	0.625	0.5298	2.9930	0.625	0.5241	0.2423	0.1906
No log	24.0	312	1.0448	0.685	0.4895	2.8196	0.685	0.5846	0.2701	0.1411
No log	25.0	325	1.0623	0.695	0.4904	2.6903	0.695	0.6086	0.2762	0.1435
No log	26.0	338	0.9872	0.695	0.4607	2.6336	0.695	0.5953	0.2728	0.1180
No log	27.0	351	0.9789	0.705	0.4580	2.6326	0.705	0.6127	0.2579	0.1171
No log	28.0	364	1.0033	0.685	0.4707	2.5747	0.685	0.5906	0.2747	0.1291
No log	29.0	377	1.0152	0.7	0.4789	2.4333	0.7	0.6260	0.2951	0.1739
No log	30.0	390	1.0107	0.715	0.4684	2.5194	0.715	0.6401	0.3197	0.1389
No log	31.0	403	0.9511	0.69	0.4445	2.5648	0.69	0.6131	0.2648	0.1298
No log	32.0	416	0.9586	0.735	0.4448	2.3342	0.735	0.6578	0.2941	0.1275
No log	33.0	429	1.0010	0.73	0.4625	2.4748	0.7300	0.6613	0.3307	0.1202
No log	34.0	442	0.9481	0.71	0.4361	2.4986	0.7100	0.6456	0.2856	0.1228
No log	35.0	455	0.9190	0.69	0.4323	2.6586	0.69	0.6265	0.2538	0.1250
No log	36.0	468	0.9226	0.715	0.4350	2.2652	0.715	0.6507	0.2868	0.1328
No log	37.0	481	0.9017	0.725	0.4182	2.5141	0.7250	0.6590	0.2547	0.1013
No log	38.0	494	0.9092	0.72	0.4218	2.5171	0.72	0.6495	0.2677	0.1055
1.0958	39.0	507	0.9093	0.71	0.4221	2.6479	0.7100	0.6456	0.2567	0.1185
1.0958	40.0	520	0.8926	0.71	0.4204	2.3785	0.7100	0.6522	0.2396	0.1153
1.0958	41.0	533	0.8928	0.715	0.4157	2.5719	0.715	0.6487	0.2708	0.1067
1.0958	42.0	546	0.8967	0.715	0.4247	2.6422	0.715	0.6495	0.2525	0.1174
1.0958	43.0	559	0.8773	0.695	0.4116	2.5548	0.695	0.6400	0.2491	0.1142
1.0958	44.0	572	0.8660	0.71	0.4036	2.2950	0.7100	0.6535	0.2401	0.1009
1.0958	45.0	585	0.8718	0.72	0.4057	2.4922	0.72	0.6551	0.2624	0.0998
1.0958	46.0	598	0.8737	0.7	0.4070	2.4455	0.7	0.6416	0.2360	0.1052
1.0958	47.0	611	0.8707	0.715	0.4094	2.3519	0.715	0.6494	0.2514	0.1086
1.0958	48.0	624	0.8640	0.705	0.4039	2.3765	0.705	0.6430	0.2538	0.1041
1.0958	49.0	637	0.8702	0.7	0.4066	2.5524	0.7	0.6423	0.2160	0.1080
1.0958	50.0	650	0.8672	0.71	0.4047	2.1924	0.7100	0.6463	0.2420	0.1050

Framework versions

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