resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t2.5_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: 0.6481
• Accuracy: 0.69
• Brier Loss: 0.4919
• Nll: 2.4969
• F1 Micro: 0.69
• F1 Macro: 0.6317
• Ece: 0.3029
• Aurc: 0.1260

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	1.4796	0.165	0.8965	8.4885	0.165	0.1123	0.2151	0.8341
No log	2.0	26	1.4679	0.165	0.8954	8.3391	0.165	0.1066	0.2136	0.8332
No log	3.0	39	1.4170	0.21	0.8858	6.1941	0.2100	0.0969	0.2433	0.7991
No log	4.0	52	1.3472	0.21	0.8711	6.0602	0.2100	0.0728	0.2320	0.7271
No log	5.0	65	1.2776	0.19	0.8572	6.1293	0.19	0.0537	0.2422	0.7473
No log	6.0	78	1.1840	0.245	0.8353	6.2405	0.245	0.1060	0.2810	0.6690
No log	7.0	91	1.0740	0.365	0.7936	6.3617	0.3650	0.1739	0.3136	0.3646
No log	8.0	104	1.1102	0.345	0.8081	5.8896	0.345	0.1812	0.3046	0.4292
No log	9.0	117	1.0735	0.34	0.7963	5.9970	0.34	0.1842	0.3028	0.4286
No log	10.0	130	1.1145	0.265	0.8110	5.9054	0.265	0.1300	0.2511	0.6350
No log	11.0	143	0.9981	0.325	0.7659	5.3834	0.325	0.1655	0.2790	0.4860
No log	12.0	156	1.0500	0.285	0.7898	4.9696	0.285	0.1594	0.2604	0.6636
No log	13.0	169	0.8764	0.445	0.6976	4.6456	0.445	0.2647	0.2779	0.3020
No log	14.0	182	0.9147	0.48	0.7108	4.4793	0.48	0.2942	0.3262	0.2862
No log	15.0	195	0.9776	0.38	0.7434	4.4065	0.38	0.2269	0.2938	0.5297
No log	16.0	208	0.8066	0.47	0.6494	3.9671	0.47	0.2966	0.2791	0.2907
No log	17.0	221	0.7766	0.535	0.6305	3.5250	0.535	0.3866	0.3003	0.2424
No log	18.0	234	0.8186	0.535	0.6458	3.3670	0.535	0.3792	0.3005	0.2311
No log	19.0	247	0.8156	0.52	0.6430	3.1633	0.52	0.3675	0.3072	0.2667
No log	20.0	260	0.8386	0.55	0.6462	3.2549	0.55	0.4251	0.3103	0.2703
No log	21.0	273	0.7996	0.515	0.6342	3.1396	0.515	0.3969	0.3177	0.2867
No log	22.0	286	0.8605	0.6	0.6472	3.2563	0.6	0.4717	0.3810	0.2113
No log	23.0	299	0.7138	0.595	0.5713	3.1171	0.595	0.4657	0.2773	0.2034
No log	24.0	312	0.7212	0.665	0.5740	2.9688	0.665	0.5474	0.3366	0.1754
No log	25.0	325	0.7463	0.63	0.5843	2.8998	0.63	0.5502	0.3432	0.2072
No log	26.0	338	0.7231	0.67	0.5626	3.1334	0.67	0.5564	0.3160	0.1521
No log	27.0	351	0.6913	0.68	0.5427	2.8906	0.68	0.5702	0.3354	0.1406
No log	28.0	364	0.6825	0.66	0.5342	2.8619	0.66	0.5615	0.2902	0.1625
No log	29.0	377	0.7015	0.665	0.5549	2.7315	0.665	0.5741	0.3305	0.1769
No log	30.0	390	0.6939	0.67	0.5406	2.7114	0.67	0.5720	0.3353	0.1420
No log	31.0	403	0.6836	0.69	0.5265	2.7567	0.69	0.5982	0.3216	0.1455
No log	32.0	416	0.6728	0.69	0.5211	2.6858	0.69	0.6056	0.3124	0.1453
No log	33.0	429	0.6926	0.675	0.5403	2.5815	0.675	0.6095	0.3258	0.1683
No log	34.0	442	0.6673	0.66	0.5090	2.5591	0.66	0.5722	0.2950	0.1385
No log	35.0	455	0.6811	0.675	0.5207	2.5813	0.675	0.5841	0.3324	0.1273
No log	36.0	468	0.6648	0.69	0.5119	2.5745	0.69	0.6225	0.3433	0.1320
No log	37.0	481	0.6623	0.67	0.5092	2.6134	0.67	0.6129	0.3204	0.1471
No log	38.0	494	0.6635	0.69	0.5088	2.3862	0.69	0.6192	0.3201	0.1311
0.7628	39.0	507	0.6554	0.685	0.5008	2.5849	0.685	0.6210	0.3179	0.1377
0.7628	40.0	520	0.6567	0.685	0.5022	2.6498	0.685	0.6310	0.3127	0.1414
0.7628	41.0	533	0.6558	0.695	0.4996	2.5917	0.695	0.6347	0.3115	0.1321
0.7628	42.0	546	0.6578	0.695	0.5021	2.4864	0.695	0.6259	0.3098	0.1306
0.7628	43.0	559	0.6544	0.685	0.4969	2.5757	0.685	0.6175	0.2955	0.1342
0.7628	44.0	572	0.6507	0.685	0.4944	2.5057	0.685	0.6257	0.3144	0.1304
0.7628	45.0	585	0.6501	0.675	0.4937	2.4903	0.675	0.6208	0.3091	0.1301
0.7628	46.0	598	0.6518	0.685	0.4949	2.4732	0.685	0.6254	0.3164	0.1235
0.7628	47.0	611	0.6499	0.685	0.4936	2.4924	0.685	0.6273	0.3124	0.1323
0.7628	48.0	624	0.6490	0.7	0.4925	2.4999	0.7	0.6353	0.3147	0.1243
0.7628	49.0	637	0.6510	0.685	0.4933	2.5758	0.685	0.6242	0.3206	0.1281
0.7628	50.0	650	0.6481	0.69	0.4919	2.4969	0.69	0.6317	0.3029	0.1260

Framework versions

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