resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t2.5_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: 0.8012
• Accuracy: 0.7
• Brier Loss: 0.4467
• Nll: 2.5682
• F1 Micro: 0.7
• F1 Macro: 0.6313
• Ece: 0.2684
• Aurc: 0.1170

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	1.8024	0.16	0.8966	8.5001	0.16	0.1073	0.2079	0.8334
No log	2.0	26	1.7941	0.145	0.8957	8.3207	0.145	0.0843	0.2022	0.8435
No log	3.0	39	1.7486	0.2	0.8868	6.2015	0.2000	0.1007	0.2209	0.7900
No log	4.0	52	1.6854	0.205	0.8738	6.0142	0.205	0.0707	0.2453	0.7584
No log	5.0	65	1.6162	0.2	0.8594	6.2364	0.2000	0.0552	0.2466	0.7717
No log	6.0	78	1.5412	0.235	0.8416	6.0423	0.235	0.0902	0.2589	0.7006
No log	7.0	91	1.5011	0.295	0.8304	6.1420	0.295	0.1272	0.2803	0.6124
No log	8.0	104	1.4415	0.3	0.8114	6.0440	0.3	0.1296	0.2870	0.5641
No log	9.0	117	1.3257	0.38	0.7625	5.6923	0.38	0.2198	0.3136	0.3675
No log	10.0	130	1.3748	0.33	0.7905	5.5276	0.33	0.1870	0.2947	0.5985
No log	11.0	143	1.3294	0.39	0.7683	4.9632	0.39	0.2573	0.2940	0.4639
No log	12.0	156	1.2444	0.385	0.7297	4.8431	0.3850	0.2330	0.2849	0.4173
No log	13.0	169	1.2212	0.45	0.7153	4.5819	0.45	0.3051	0.3143	0.3379
No log	14.0	182	1.1835	0.495	0.6888	3.6108	0.495	0.3412	0.3316	0.2873
No log	15.0	195	1.1203	0.47	0.6559	3.6500	0.47	0.3348	0.2935	0.3061
No log	16.0	208	1.1520	0.495	0.6707	3.8106	0.495	0.3632	0.2938	0.3604
No log	17.0	221	1.0261	0.565	0.6021	3.3382	0.565	0.4214	0.2840	0.2047
No log	18.0	234	1.0080	0.61	0.5914	3.2936	0.61	0.4748	0.3240	0.1806
No log	19.0	247	1.0696	0.58	0.6253	3.2354	0.58	0.4686	0.3152	0.2626
No log	20.0	260	0.9733	0.615	0.5722	3.1019	0.615	0.4968	0.3259	0.2066
No log	21.0	273	0.9266	0.625	0.5423	3.0239	0.625	0.5202	0.2834	0.1782
No log	22.0	286	0.9364	0.66	0.5461	2.9031	0.66	0.5461	0.3128	0.1601
No log	23.0	299	0.9181	0.675	0.5307	2.8416	0.675	0.5584	0.3106	0.1462
No log	24.0	312	0.9739	0.665	0.5539	2.8798	0.665	0.5634	0.3325	0.1610
No log	25.0	325	0.8851	0.69	0.5099	2.7336	0.69	0.6013	0.3064	0.1437
No log	26.0	338	0.8755	0.71	0.4979	2.7400	0.7100	0.6032	0.3162	0.1211
No log	27.0	351	0.8653	0.675	0.4964	2.8339	0.675	0.5705	0.2977	0.1386
No log	28.0	364	0.8838	0.675	0.5055	2.7456	0.675	0.5816	0.2969	0.1524
No log	29.0	377	0.8805	0.68	0.5025	2.6942	0.68	0.5855	0.3099	0.1380
No log	30.0	390	0.8585	0.665	0.4891	2.7511	0.665	0.5737	0.2627	0.1370
No log	31.0	403	0.8410	0.675	0.4736	2.6431	0.675	0.5985	0.2670	0.1335
No log	32.0	416	0.8378	0.71	0.4724	2.7320	0.7100	0.6236	0.2885	0.1153
No log	33.0	429	0.8421	0.705	0.4718	2.6331	0.705	0.6326	0.2644	0.1147
No log	34.0	442	0.8350	0.685	0.4697	2.8035	0.685	0.6062	0.2831	0.1291
No log	35.0	455	0.8377	0.7	0.4708	2.4611	0.7	0.6376	0.3173	0.1195
No log	36.0	468	0.8126	0.69	0.4562	2.3909	0.69	0.6154	0.2433	0.1177
No log	37.0	481	0.8299	0.685	0.4673	2.5695	0.685	0.6080	0.2802	0.1261
No log	38.0	494	0.8197	0.685	0.4597	2.6388	0.685	0.6187	0.2690	0.1229
0.9314	39.0	507	0.8137	0.695	0.4547	2.7263	0.695	0.6332	0.2581	0.1207
0.9314	40.0	520	0.8168	0.69	0.4583	2.6230	0.69	0.6267	0.2696	0.1161
0.9314	41.0	533	0.8090	0.7	0.4529	2.6449	0.7	0.6236	0.2445	0.1187
0.9314	42.0	546	0.8168	0.68	0.4586	2.5516	0.68	0.6162	0.2722	0.1275
0.9314	43.0	559	0.8100	0.7	0.4523	2.5565	0.7	0.6347	0.2869	0.1192
0.9314	44.0	572	0.8078	0.7	0.4514	2.5734	0.7	0.6344	0.2583	0.1172
0.9314	45.0	585	0.8022	0.715	0.4472	2.4971	0.715	0.6534	0.2890	0.1165
0.9314	46.0	598	0.8049	0.695	0.4484	2.4891	0.695	0.6423	0.2722	0.1189
0.9314	47.0	611	0.8025	0.705	0.4481	2.4929	0.705	0.6393	0.2650	0.1124
0.9314	48.0	624	0.7973	0.7	0.4439	2.5000	0.7	0.6292	0.2718	0.1142
0.9314	49.0	637	0.8011	0.7	0.4464	2.5713	0.7	0.6303	0.2400	0.1183
0.9314	50.0	650	0.8012	0.7	0.4467	2.5682	0.7	0.6313	0.2684	0.1170

Framework versions

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