resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t5.0_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: 0.7861
• Accuracy: 0.705
• Brier Loss: 0.4410
• Nll: 2.6519
• F1 Micro: 0.705
• F1 Macro: 0.6403
• Ece: 0.2724
• Aurc: 0.1188

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	1.7831	0.165	0.8966	8.4414	0.165	0.1121	0.2151	0.8335
No log	2.0	26	1.7753	0.145	0.8958	8.5715	0.145	0.0954	0.1998	0.8332
No log	3.0	39	1.7334	0.175	0.8877	6.4682	0.175	0.0756	0.2069	0.7896
No log	4.0	52	1.6604	0.185	0.8723	6.0351	0.185	0.0505	0.2328	0.7549
No log	5.0	65	1.5874	0.18	0.8560	6.0732	0.18	0.0431	0.2285	0.7506
No log	6.0	78	1.5223	0.185	0.8415	6.1638	0.185	0.0479	0.2419	0.7530
No log	7.0	91	1.4642	0.35	0.8239	6.0328	0.35	0.1696	0.3081	0.5219
No log	8.0	104	1.3599	0.35	0.7825	6.2102	0.35	0.1908	0.2977	0.4172
No log	9.0	117	1.3083	0.385	0.7566	5.7128	0.3850	0.2203	0.3012	0.3842
No log	10.0	130	1.3151	0.365	0.7670	5.1073	0.3650	0.2150	0.2923	0.4891
No log	11.0	143	1.3736	0.295	0.7950	5.3584	0.295	0.1747	0.2716	0.6360
No log	12.0	156	1.2655	0.425	0.7380	4.0312	0.425	0.2789	0.3273	0.3366
No log	13.0	169	1.1696	0.475	0.6901	3.9627	0.4750	0.3083	0.3011	0.2825
No log	14.0	182	1.2992	0.355	0.7473	3.9098	0.3550	0.2292	0.2675	0.4929
No log	15.0	195	1.1698	0.51	0.6881	3.7143	0.51	0.3691	0.3333	0.3278
No log	16.0	208	1.0624	0.515	0.6274	3.8387	0.515	0.3631	0.2821	0.2583
No log	17.0	221	1.0970	0.565	0.6421	3.3302	0.565	0.4493	0.3362	0.2373
No log	18.0	234	1.0029	0.625	0.5883	3.3820	0.625	0.4675	0.3005	0.1660
No log	19.0	247	1.0384	0.605	0.6093	3.3183	0.605	0.4863	0.3252	0.2145
No log	20.0	260	1.0686	0.62	0.6234	3.0246	0.62	0.5155	0.3625	0.2334
No log	21.0	273	0.9641	0.62	0.5685	2.9225	0.62	0.5259	0.3103	0.2063
No log	22.0	286	1.0054	0.665	0.5849	3.0792	0.665	0.5614	0.3636	0.1863
No log	23.0	299	0.9959	0.675	0.5734	2.9829	0.675	0.5577	0.3619	0.1806
No log	24.0	312	0.9044	0.675	0.5267	2.8952	0.675	0.5712	0.2989	0.1475
No log	25.0	325	0.9803	0.655	0.5627	2.7501	0.655	0.5418	0.3415	0.1919
No log	26.0	338	0.8814	0.65	0.5176	2.8421	0.65	0.5619	0.2665	0.1694
No log	27.0	351	0.8555	0.69	0.4928	2.7870	0.69	0.5831	0.3091	0.1279
No log	28.0	364	0.8290	0.69	0.4777	2.6377	0.69	0.5976	0.2551	0.1290
No log	29.0	377	0.8593	0.685	0.4949	2.5880	0.685	0.5776	0.3083	0.1279
No log	30.0	390	0.8226	0.685	0.4678	2.8938	0.685	0.5884	0.2820	0.1249
No log	31.0	403	0.8578	0.69	0.4857	2.6150	0.69	0.6024	0.3109	0.1344
No log	32.0	416	0.8330	0.685	0.4753	2.5999	0.685	0.6047	0.2688	0.1407
No log	33.0	429	0.8268	0.7	0.4683	2.6138	0.7	0.6193	0.2913	0.1315
No log	34.0	442	0.8535	0.715	0.4749	2.5059	0.715	0.6450	0.2931	0.1190
No log	35.0	455	0.8334	0.665	0.4752	2.3839	0.665	0.5950	0.2762	0.1397
No log	36.0	468	0.8025	0.71	0.4553	2.4803	0.7100	0.6302	0.2889	0.1178
No log	37.0	481	0.8142	0.715	0.4563	2.6785	0.715	0.6426	0.2989	0.1048
No log	38.0	494	0.8124	0.7	0.4538	2.5320	0.7	0.6332	0.2594	0.1132
0.9303	39.0	507	0.7888	0.69	0.4452	2.6427	0.69	0.6269	0.2583	0.1224
0.9303	40.0	520	0.7907	0.705	0.4458	2.6942	0.705	0.6367	0.2688	0.1155
0.9303	41.0	533	0.7918	0.71	0.4442	2.4378	0.7100	0.6558	0.2816	0.1132
0.9303	42.0	546	0.8005	0.725	0.4479	2.6088	0.7250	0.6576	0.2914	0.1049
0.9303	43.0	559	0.7879	0.72	0.4421	2.7052	0.72	0.6592	0.2741	0.1122
0.9303	44.0	572	0.7910	0.71	0.4461	2.6463	0.7100	0.6463	0.3119	0.1188
0.9303	45.0	585	0.7922	0.705	0.4450	2.6453	0.705	0.6481	0.2753	0.1211
0.9303	46.0	598	0.7915	0.715	0.4429	2.6970	0.715	0.6526	0.2741	0.1107
0.9303	47.0	611	0.7809	0.705	0.4370	2.6841	0.705	0.6453	0.2734	0.1158
0.9303	48.0	624	0.7771	0.705	0.4350	2.6168	0.705	0.6423	0.2652	0.1139
0.9303	49.0	637	0.7826	0.705	0.4377	2.5091	0.705	0.6423	0.2758	0.1202
0.9303	50.0	650	0.7861	0.705	0.4410	2.6519	0.705	0.6403	0.2724	0.1188

Framework versions

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