resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t1.5_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: 0.6500
• Accuracy: 0.69
• Brier Loss: 0.5003
• Nll: 2.5629
• F1 Micro: 0.69
• F1 Macro: 0.6350
• Ece: 0.3098
• Aurc: 0.1329

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	1.4712	0.165	0.8966	8.4652	0.165	0.1101	0.2129	0.8342
No log	2.0	26	1.4590	0.165	0.8951	8.1097	0.165	0.1059	0.2034	0.8021
No log	3.0	39	1.4178	0.175	0.8873	6.8095	0.175	0.0813	0.2150	0.7994
No log	4.0	52	1.3342	0.18	0.8702	6.4137	0.18	0.0475	0.2314	0.7558
No log	5.0	65	1.2828	0.2	0.8587	6.1547	0.2000	0.0642	0.2429	0.7009
No log	6.0	78	1.2675	0.205	0.8548	6.1395	0.205	0.0612	0.2348	0.7022
No log	7.0	91	1.0716	0.31	0.7962	6.4589	0.31	0.1241	0.2787	0.4433
No log	8.0	104	1.1184	0.29	0.8126	6.2585	0.29	0.1394	0.2863	0.5819
No log	9.0	117	1.1021	0.31	0.8075	6.0370	0.31	0.1697	0.2834	0.5458
No log	10.0	130	1.0268	0.33	0.7815	6.1370	0.33	0.1921	0.2856	0.5395
No log	11.0	143	1.0290	0.355	0.7759	5.3640	0.3550	0.2143	0.2795	0.4697
No log	12.0	156	0.9169	0.36	0.7262	5.2997	0.36	0.1995	0.2761	0.4070
No log	13.0	169	0.9903	0.36	0.7586	4.9404	0.36	0.2200	0.2832	0.5343
No log	14.0	182	0.9128	0.425	0.7082	4.5862	0.425	0.2706	0.2834	0.3542
No log	15.0	195	1.0046	0.405	0.7441	3.9763	0.405	0.2759	0.3142	0.4602
No log	16.0	208	0.9277	0.41	0.7146	4.3670	0.41	0.2763	0.2695	0.4409
No log	17.0	221	0.9726	0.505	0.7208	3.5350	0.505	0.3736	0.3332	0.3469
No log	18.0	234	0.7717	0.505	0.6280	3.4386	0.505	0.3412	0.2564	0.2567
No log	19.0	247	0.7723	0.58	0.6143	3.6207	0.58	0.4125	0.3178	0.1847
No log	20.0	260	0.8182	0.57	0.6419	3.1633	0.57	0.4855	0.3517	0.2530
No log	21.0	273	0.7333	0.58	0.5891	3.3014	0.58	0.4512	0.2718	0.2137
No log	22.0	286	0.7374	0.665	0.5856	3.0299	0.665	0.5432	0.3459	0.1657
No log	23.0	299	0.7083	0.645	0.5564	3.0874	0.645	0.5180	0.3112	0.1608
No log	24.0	312	0.7480	0.64	0.5901	3.0218	0.64	0.5410	0.3701	0.1976
No log	25.0	325	0.7547	0.68	0.5894	2.9002	0.68	0.5801	0.3817	0.1559
No log	26.0	338	0.6998	0.65	0.5474	2.9402	0.65	0.5468	0.2875	0.1707
No log	27.0	351	0.6967	0.66	0.5506	2.8344	0.66	0.5578	0.3105	0.1707
No log	28.0	364	0.6733	0.655	0.5332	2.6492	0.655	0.5719	0.2935	0.1554
No log	29.0	377	0.7162	0.67	0.5596	2.7250	0.67	0.5721	0.3388	0.1423
No log	30.0	390	0.6826	0.665	0.5291	2.7460	0.665	0.5797	0.3353	0.1469
No log	31.0	403	0.6761	0.665	0.5195	2.7938	0.665	0.5647	0.3096	0.1485
No log	32.0	416	0.6745	0.695	0.5295	2.6172	0.695	0.6160	0.3171	0.1636
No log	33.0	429	0.6785	0.695	0.5242	2.5816	0.695	0.6115	0.3475	0.1349
No log	34.0	442	0.6688	0.665	0.5174	2.6401	0.665	0.5833	0.2988	0.1427
No log	35.0	455	0.6767	0.675	0.5275	2.6364	0.675	0.6027	0.3285	0.1483
No log	36.0	468	0.6605	0.695	0.5076	2.6483	0.695	0.6252	0.3127	0.1372
No log	37.0	481	0.6538	0.705	0.5029	2.6284	0.705	0.6340	0.3173	0.1220
No log	38.0	494	0.6610	0.695	0.5102	2.5052	0.695	0.6375	0.3128	0.1298
0.7532	39.0	507	0.6618	0.695	0.5110	2.5663	0.695	0.6268	0.3297	0.1367
0.7532	40.0	520	0.6749	0.69	0.5235	2.5343	0.69	0.6341	0.3256	0.1332
0.7532	41.0	533	0.6574	0.695	0.5062	2.4223	0.695	0.6338	0.3292	0.1469
0.7532	42.0	546	0.6530	0.695	0.5026	2.6189	0.695	0.6390	0.2950	0.1391
0.7532	43.0	559	0.6509	0.685	0.5003	2.5417	0.685	0.6299	0.3150	0.1368
0.7532	44.0	572	0.6520	0.71	0.5030	2.4796	0.7100	0.6453	0.3251	0.1286
0.7532	45.0	585	0.6494	0.69	0.4994	2.5431	0.69	0.6327	0.3138	0.1279
0.7532	46.0	598	0.6515	0.71	0.5007	2.5295	0.7100	0.6541	0.3307	0.1208
0.7532	47.0	611	0.6477	0.69	0.4979	2.5971	0.69	0.6323	0.3263	0.1281
0.7532	48.0	624	0.6495	0.7	0.5007	2.6162	0.7	0.6395	0.3412	0.1272
0.7532	49.0	637	0.6478	0.7	0.4968	2.4946	0.7	0.6386	0.3191	0.1309
0.7532	50.0	650	0.6500	0.69	0.5003	2.5629	0.69	0.6350	0.3098	0.1329

Framework versions

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