resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t1.5_a0.7

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: 0.8009
• Accuracy: 0.695
• Brier Loss: 0.4518
• Nll: 2.3840
• F1 Micro: 0.695
• F1 Macro: 0.6406
• Ece: 0.2661
• Aurc: 0.1211

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	1.7971	0.17	0.8966	8.4593	0.17	0.1148	0.2202	0.8308
No log	2.0	26	1.7887	0.13	0.8956	8.3211	0.13	0.0772	0.2024	0.8359
No log	3.0	39	1.7450	0.225	0.8868	6.4554	0.225	0.1165	0.2502	0.7572
No log	4.0	52	1.6811	0.24	0.8733	5.9510	0.24	0.0953	0.2651	0.6944
No log	5.0	65	1.6411	0.19	0.8649	6.0993	0.19	0.0493	0.2422	0.7783
No log	6.0	78	1.5475	0.195	0.8429	6.2065	0.195	0.0630	0.2472	0.7110
No log	7.0	91	1.4688	0.3	0.8197	6.0345	0.3	0.1481	0.2936	0.5379
No log	8.0	104	1.5036	0.285	0.8294	5.6660	0.285	0.1428	0.2869	0.6535
No log	9.0	117	1.3901	0.34	0.7934	5.9107	0.34	0.1853	0.2894	0.5277
No log	10.0	130	1.3484	0.37	0.7760	5.6441	0.37	0.2175	0.3177	0.5266
No log	11.0	143	1.3375	0.34	0.7734	5.0872	0.34	0.2083	0.2902	0.5557
No log	12.0	156	1.3639	0.305	0.7834	4.5070	0.305	0.1885	0.2674	0.6177
No log	13.0	169	1.2321	0.415	0.7225	4.3464	0.415	0.2751	0.2943	0.3825
No log	14.0	182	1.1453	0.44	0.6767	4.4158	0.44	0.2864	0.2617	0.3413
No log	15.0	195	1.1830	0.43	0.6965	3.8251	0.4300	0.2972	0.2912	0.4239
No log	16.0	208	1.0572	0.535	0.6230	3.5943	0.535	0.3758	0.2861	0.2291
No log	17.0	221	1.0532	0.585	0.6151	3.3834	0.585	0.4331	0.3278	0.1879
No log	18.0	234	1.0940	0.565	0.6374	3.2290	0.565	0.4431	0.3313	0.2415
No log	19.0	247	0.9877	0.585	0.5886	3.1068	0.585	0.4564	0.2896	0.2110
No log	20.0	260	1.0405	0.61	0.6056	3.1786	0.61	0.5038	0.3428	0.1962
No log	21.0	273	0.9728	0.635	0.5634	2.9133	0.635	0.5293	0.3333	0.1664
No log	22.0	286	0.9425	0.635	0.5527	2.8909	0.635	0.5237	0.3131	0.1796
No log	23.0	299	0.9549	0.65	0.5605	2.8074	0.65	0.5539	0.3283	0.1914
No log	24.0	312	1.0085	0.67	0.5733	2.8377	0.67	0.5543	0.3525	0.1571
No log	25.0	325	0.9140	0.655	0.5257	2.5878	0.655	0.5603	0.3171	0.1495
No log	26.0	338	0.8979	0.65	0.5249	2.7723	0.65	0.5563	0.2843	0.1646
No log	27.0	351	0.8912	0.675	0.5082	2.6562	0.675	0.5837	0.2871	0.1380
No log	28.0	364	0.8966	0.66	0.5242	2.3150	0.66	0.5890	0.3180	0.1777
No log	29.0	377	0.8602	0.67	0.4959	2.5813	0.67	0.5866	0.3023	0.1319
No log	30.0	390	0.8434	0.69	0.4779	2.5451	0.69	0.6130	0.3061	0.1188
No log	31.0	403	0.8406	0.715	0.4782	2.3339	0.715	0.6438	0.3241	0.1092
No log	32.0	416	0.8294	0.71	0.4726	2.5394	0.7100	0.6308	0.2922	0.1218
No log	33.0	429	0.8329	0.68	0.4763	2.4520	0.68	0.6166	0.2592	0.1396
No log	34.0	442	0.8937	0.69	0.5015	2.5649	0.69	0.6357	0.3293	0.1279
No log	35.0	455	0.8358	0.665	0.4807	2.4437	0.665	0.6178	0.2380	0.1473
No log	36.0	468	0.8283	0.685	0.4747	2.5408	0.685	0.6304	0.3126	0.1361
No log	37.0	481	0.8235	0.685	0.4707	2.4620	0.685	0.6300	0.2757	0.1343
No log	38.0	494	0.8289	0.68	0.4778	2.5443	0.68	0.6305	0.2935	0.1469
0.9462	39.0	507	0.8373	0.69	0.4728	2.5775	0.69	0.6281	0.3028	0.1149
0.9462	40.0	520	0.8062	0.715	0.4548	2.3673	0.715	0.6587	0.2776	0.1133
0.9462	41.0	533	0.7990	0.705	0.4517	2.3284	0.705	0.6463	0.2716	0.1185
0.9462	42.0	546	0.8210	0.7	0.4650	2.5646	0.7	0.6432	0.2690	0.1199
0.9462	43.0	559	0.8102	0.695	0.4558	2.5651	0.695	0.6442	0.2656	0.1184
0.9462	44.0	572	0.8061	0.69	0.4566	2.5154	0.69	0.6356	0.2816	0.1267
0.9462	45.0	585	0.8018	0.7	0.4531	2.4982	0.7	0.6419	0.2696	0.1192
0.9462	46.0	598	0.8040	0.7	0.4521	2.5309	0.7	0.6448	0.2797	0.1166
0.9462	47.0	611	0.8062	0.68	0.4560	2.5452	0.68	0.6370	0.2744	0.1217
0.9462	48.0	624	0.8011	0.69	0.4529	2.4281	0.69	0.6402	0.2594	0.1224
0.9462	49.0	637	0.8017	0.69	0.4532	2.4239	0.69	0.6400	0.2613	0.1261
0.9462	50.0	650	0.8009	0.695	0.4518	2.3840	0.695	0.6406	0.2661	0.1211

Framework versions

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