resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t1.5_a0.9

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: 0.8831
• Accuracy: 0.695
• Brier Loss: 0.4126
• Nll: 2.4628
• F1 Micro: 0.695
• F1 Macro: 0.6387
• Ece: 0.2426
• Aurc: 0.1068

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	2.1233	0.16	0.8967	8.5697	0.16	0.1066	0.2078	0.8316
No log	2.0	26	2.1188	0.14	0.8961	8.2960	0.14	0.0886	0.1947	0.8419
No log	3.0	39	2.0764	0.195	0.8873	6.4713	0.195	0.1159	0.2335	0.7665
No log	4.0	52	2.0365	0.21	0.8787	5.7752	0.2100	0.0930	0.2376	0.7548
No log	5.0	65	1.9888	0.2	0.8682	5.8737	0.2000	0.0775	0.2417	0.7314
No log	6.0	78	1.8998	0.215	0.8465	5.8553	0.2150	0.0970	0.2586	0.7063
No log	7.0	91	1.8351	0.33	0.8289	5.7781	0.33	0.1904	0.3089	0.6103
No log	8.0	104	1.7342	0.4	0.7968	5.5366	0.4000	0.2476	0.3457	0.4276
No log	9.0	117	1.6787	0.36	0.7757	5.7414	0.36	0.2148	0.3062	0.4324
No log	10.0	130	1.6942	0.4	0.7870	5.2615	0.4000	0.2831	0.3168	0.5227
No log	11.0	143	1.5992	0.4	0.7489	4.7833	0.4000	0.2649	0.3053	0.4679
No log	12.0	156	1.6071	0.425	0.7532	4.2803	0.425	0.2906	0.3196	0.4646
No log	13.0	169	1.4727	0.48	0.6925	4.1911	0.48	0.3239	0.2957	0.3081
No log	14.0	182	1.4275	0.515	0.6705	3.7980	0.515	0.3569	0.3211	0.2626
No log	15.0	195	1.3282	0.56	0.6200	3.6359	0.56	0.4163	0.2990	0.2213
No log	16.0	208	1.3280	0.565	0.6263	3.4960	0.565	0.4177	0.3217	0.2346
No log	17.0	221	1.3220	0.595	0.6196	3.2202	0.595	0.4639	0.3322	0.1992
No log	18.0	234	1.2359	0.595	0.5840	3.3332	0.595	0.4780	0.3042	0.2011
No log	19.0	247	1.1690	0.625	0.5531	3.2423	0.625	0.5233	0.2940	0.1807
No log	20.0	260	1.1644	0.64	0.5532	3.0542	0.64	0.5429	0.3019	0.1821
No log	21.0	273	1.1611	0.62	0.5516	2.9412	0.62	0.5193	0.2865	0.2160
No log	22.0	286	1.3427	0.585	0.6361	3.0936	0.585	0.5089	0.3442	0.2922
No log	23.0	299	1.1238	0.62	0.5440	2.7924	0.62	0.5458	0.2654	0.2088
No log	24.0	312	1.2008	0.685	0.5615	2.5918	0.685	0.5890	0.3907	0.1516
No log	25.0	325	1.0764	0.695	0.5000	2.6354	0.695	0.6107	0.3126	0.1397
No log	26.0	338	1.0268	0.675	0.4822	2.4798	0.675	0.5992	0.2775	0.1229
No log	27.0	351	1.0340	0.67	0.4893	2.4316	0.67	0.5997	0.2763	0.1638
No log	28.0	364	1.0154	0.665	0.4769	2.6487	0.665	0.6034	0.2590	0.1487
No log	29.0	377	1.0013	0.64	0.4814	2.5899	0.64	0.5771	0.2429	0.1593
No log	30.0	390	1.0173	0.685	0.4714	2.6922	0.685	0.6178	0.2898	0.1423
No log	31.0	403	0.9630	0.695	0.4509	2.6349	0.695	0.6206	0.2746	0.1248
No log	32.0	416	0.9950	0.68	0.4648	2.4144	0.68	0.6362	0.3020	0.1725
No log	33.0	429	0.9711	0.72	0.4502	2.6651	0.72	0.6571	0.2892	0.1268
No log	34.0	442	0.9491	0.705	0.4425	2.7169	0.705	0.6425	0.2541	0.1145
No log	35.0	455	0.9213	0.685	0.4309	2.5736	0.685	0.6174	0.2380	0.1161
No log	36.0	468	0.9144	0.695	0.4297	2.4141	0.695	0.6308	0.2502	0.1154
No log	37.0	481	0.9242	0.715	0.4264	2.7191	0.715	0.6429	0.2386	0.1030
No log	38.0	494	0.9290	0.695	0.4346	2.6515	0.695	0.6367	0.2432	0.1189
1.0953	39.0	507	0.9110	0.69	0.4262	2.6615	0.69	0.6328	0.2368	0.1112
1.0953	40.0	520	0.9000	0.695	0.4186	2.4590	0.695	0.6417	0.2453	0.1070
1.0953	41.0	533	0.8961	0.69	0.4189	2.4170	0.69	0.6368	0.2349	0.1090
1.0953	42.0	546	0.9103	0.675	0.4286	2.6129	0.675	0.6193	0.2318	0.1190
1.0953	43.0	559	0.8858	0.715	0.4131	2.5243	0.715	0.6517	0.2462	0.1053
1.0953	44.0	572	0.8872	0.705	0.4135	2.3272	0.705	0.6542	0.2596	0.1051
1.0953	45.0	585	0.8897	0.715	0.4136	2.3788	0.715	0.6532	0.2560	0.1035
1.0953	46.0	598	0.8842	0.7	0.4117	2.5325	0.7	0.6446	0.2327	0.1075
1.0953	47.0	611	0.8857	0.675	0.4141	2.5451	0.675	0.6203	0.2473	0.1125
1.0953	48.0	624	0.8875	0.69	0.4164	2.4696	0.69	0.6352	0.2542	0.1109
1.0953	49.0	637	0.8842	0.69	0.4153	2.5338	0.69	0.6358	0.2302	0.1112
1.0953	50.0	650	0.8831	0.695	0.4126	2.4628	0.695	0.6387	0.2426	0.1068

Framework versions

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