resnet101-base_tobacco-cnn_tobacco3482_kd

This model is a fine-tuned version of bdpc/resnet101-base_tobacco on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8158
• Accuracy: 0.565
• Brier Loss: 0.6104
• Nll: 2.6027
• F1 Micro: 0.565
• F1 Macro: 0.4783
• Ece: 0.2677
• Aurc: 0.2516

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: 256
• eval_batch_size: 256
• 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	4	1.4988	0.065	0.9007	9.6504	0.065	0.0267	0.1512	0.9377
No log	2.0	8	1.4615	0.155	0.8961	7.9200	0.155	0.0268	0.2328	0.9605
No log	3.0	12	1.4913	0.155	0.9531	11.6402	0.155	0.0268	0.3390	0.8899
No log	4.0	16	2.2747	0.155	1.4111	11.0294	0.155	0.0268	0.7077	0.7068
No log	5.0	20	2.4543	0.155	1.4359	8.3074	0.155	0.0268	0.7226	0.6151
No log	6.0	24	1.9614	0.155	1.1785	6.6431	0.155	0.0283	0.5497	0.6022
No log	7.0	28	1.6280	0.18	0.9978	5.8468	0.18	0.0488	0.4014	0.6135
No log	8.0	32	1.3465	0.225	0.8993	5.6177	0.225	0.0740	0.3378	0.5786
No log	9.0	36	1.2597	0.225	0.8794	5.0542	0.225	0.0727	0.3403	0.5658
No log	10.0	40	1.1149	0.27	0.8181	4.5188	0.27	0.1222	0.2890	0.5230
No log	11.0	44	0.9805	0.31	0.7600	3.8687	0.31	0.1726	0.2703	0.4690
No log	12.0	48	1.0099	0.335	0.7732	3.6652	0.335	0.2095	0.2892	0.4739
No log	13.0	52	1.0522	0.335	0.7919	3.3843	0.335	0.2562	0.3006	0.6402
No log	14.0	56	1.0566	0.32	0.7868	3.4244	0.32	0.2373	0.3023	0.6094
No log	15.0	60	0.9670	0.405	0.7333	3.3926	0.405	0.3189	0.3013	0.4037
No log	16.0	64	1.0979	0.31	0.7877	3.3045	0.31	0.2262	0.2792	0.5720
No log	17.0	68	0.9022	0.44	0.6913	3.2277	0.44	0.3429	0.2902	0.3657
No log	18.0	72	1.2120	0.315	0.8075	4.1289	0.315	0.2323	0.2857	0.5909
No log	19.0	76	1.1945	0.39	0.7974	4.2350	0.39	0.3292	0.3271	0.5989
No log	20.0	80	1.3861	0.345	0.7981	5.2605	0.345	0.2700	0.2832	0.5299
No log	21.0	84	1.2243	0.33	0.8073	4.5262	0.33	0.2545	0.3068	0.6133
No log	22.0	88	1.0455	0.38	0.7238	2.7133	0.38	0.3084	0.2901	0.4855
No log	23.0	92	0.9044	0.45	0.6814	3.4361	0.45	0.3273	0.2927	0.3246
No log	24.0	96	0.8930	0.495	0.6596	3.3412	0.495	0.4185	0.2882	0.3070
No log	25.0	100	0.8665	0.485	0.6534	2.9998	0.485	0.4154	0.2641	0.3298
No log	26.0	104	1.0458	0.375	0.7579	3.1074	0.375	0.3333	0.2735	0.5293
No log	27.0	108	1.0170	0.41	0.7321	2.8884	0.41	0.3468	0.2976	0.4566
No log	28.0	112	1.0956	0.395	0.7464	3.3094	0.395	0.3255	0.3154	0.4684
No log	29.0	116	1.0805	0.39	0.7544	3.2115	0.39	0.3193	0.3014	0.4594
No log	30.0	120	1.2358	0.375	0.7733	4.3992	0.375	0.3058	0.2845	0.4876
No log	31.0	124	1.0532	0.4	0.7458	2.7398	0.4000	0.3614	0.2890	0.4961
No log	32.0	128	1.0166	0.365	0.7355	2.5093	0.3650	0.2862	0.2728	0.5057
No log	33.0	132	0.9395	0.48	0.6807	2.6211	0.48	0.4394	0.2843	0.3719
No log	34.0	136	0.8718	0.52	0.6538	2.6802	0.52	0.4697	0.2954	0.3051
No log	35.0	140	0.8339	0.51	0.6362	3.1084	0.51	0.4373	0.2654	0.3006
No log	36.0	144	0.8411	0.51	0.6359	2.7881	0.51	0.4286	0.2759	0.2906
No log	37.0	148	0.8556	0.505	0.6402	2.5519	0.505	0.4076	0.2522	0.3060
No log	38.0	152	1.0928	0.395	0.7438	2.8660	0.395	0.3337	0.2815	0.4724
No log	39.0	156	1.3830	0.39	0.8135	4.7392	0.39	0.3094	0.2879	0.5239
No log	40.0	160	1.2180	0.38	0.7760	3.8384	0.38	0.3106	0.2614	0.5109
No log	41.0	164	1.1337	0.365	0.7486	2.8843	0.3650	0.2948	0.2665	0.4630
No log	42.0	168	0.8814	0.53	0.6425	2.3353	0.53	0.4645	0.2968	0.2973
No log	43.0	172	0.8324	0.515	0.6174	2.4407	0.515	0.4517	0.2847	0.2742
No log	44.0	176	0.8477	0.53	0.6282	2.5469	0.53	0.4615	0.2712	0.2831
No log	45.0	180	0.8307	0.515	0.6190	2.4871	0.515	0.4404	0.2594	0.2845
No log	46.0	184	0.8116	0.53	0.6070	2.4944	0.53	0.4410	0.2337	0.2451
No log	47.0	188	0.8349	0.54	0.6260	2.2843	0.54	0.4423	0.2911	0.2616
No log	48.0	192	0.8298	0.555	0.6178	2.2946	0.555	0.4725	0.2568	0.2482
No log	49.0	196	0.8252	0.565	0.6141	2.3311	0.565	0.4762	0.2810	0.2504
No log	50.0	200	0.8158	0.565	0.6104	2.6027	0.565	0.4783	0.2677	0.2516

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.2.0.dev20231112+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1