dit-base_tobacco_crl

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

• Loss: 0.3591
• Accuracy: 0.935
• Brier Loss: 0.1049
• Nll: 0.7583
• F1 Micro: 0.935
• F1 Macro: 0.9303
• Ece: 0.0608
• Aurc: 0.0092

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	0.96	3	0.3864	0.925	0.1155	1.0908	0.925	0.9187	0.0833	0.0112
No log	1.96	6	0.3462	0.935	0.1104	1.0903	0.935	0.9308	0.0791	0.0123
No log	2.96	9	0.3278	0.94	0.1121	1.0884	0.94	0.9399	0.0806	0.0143
No log	3.96	12	0.3634	0.925	0.1178	1.0882	0.925	0.9213	0.0849	0.0162
No log	4.96	15	0.3682	0.925	0.1187	1.0834	0.925	0.9213	0.0763	0.0167
No log	5.96	18	0.3456	0.93	0.1133	1.0735	0.93	0.9247	0.0710	0.0148
No log	6.96	21	0.3335	0.93	0.1107	1.0762	0.93	0.9236	0.0598	0.0119
No log	7.96	24	0.3303	0.925	0.1131	1.0704	0.925	0.9204	0.0713	0.0113
No log	8.96	27	0.3394	0.93	0.1110	1.0586	0.93	0.9246	0.0731	0.0115
No log	9.96	30	0.3348	0.94	0.1067	1.0420	0.94	0.9374	0.0677	0.0127
No log	10.96	33	0.3362	0.94	0.1047	1.0205	0.94	0.9375	0.0732	0.0137
No log	11.96	36	0.3314	0.94	0.1042	1.0173	0.94	0.9390	0.0679	0.0123
No log	12.96	39	0.3310	0.94	0.1089	1.0218	0.94	0.9390	0.0728	0.0116
No log	13.96	42	0.3400	0.935	0.1100	1.0094	0.935	0.9361	0.0704	0.0140
No log	14.96	45	0.3507	0.93	0.1142	0.9968	0.93	0.9292	0.0710	0.0151
No log	15.96	48	0.3661	0.93	0.1178	0.9775	0.93	0.9315	0.0668	0.0154
No log	16.96	51	0.3673	0.93	0.1149	0.9696	0.93	0.9315	0.0580	0.0135
No log	17.96	54	0.3520	0.935	0.1112	0.9773	0.935	0.9317	0.0700	0.0113
No log	18.96	57	0.3450	0.93	0.1074	0.9759	0.93	0.9273	0.0580	0.0108
No log	19.96	60	0.3381	0.94	0.0998	0.9696	0.94	0.9445	0.0587	0.0124
No log	20.96	63	0.3371	0.935	0.0995	0.9645	0.935	0.9346	0.0578	0.0132
No log	21.96	66	0.3452	0.945	0.1014	0.9606	0.945	0.9446	0.0617	0.0147
No log	22.96	69	0.3627	0.94	0.1034	0.9504	0.94	0.9416	0.0655	0.0141
No log	23.96	72	0.3576	0.945	0.1025	0.9361	0.945	0.9445	0.0583	0.0123
No log	24.96	75	0.3528	0.94	0.1008	0.9230	0.94	0.9390	0.0610	0.0116
No log	25.96	78	0.3514	0.935	0.1030	0.9219	0.935	0.9338	0.0609	0.0112
No log	26.96	81	0.3560	0.94	0.1031	0.9334	0.94	0.9370	0.0623	0.0105
No log	27.96	84	0.3613	0.945	0.1038	0.9425	0.945	0.9422	0.0570	0.0102
No log	28.96	87	0.3701	0.94	0.1065	0.9366	0.94	0.9353	0.0605	0.0098
No log	29.96	90	0.3682	0.94	0.1064	0.9261	0.94	0.9353	0.0662	0.0094
No log	30.96	93	0.3583	0.945	0.1022	0.9012	0.945	0.9422	0.0591	0.0100
No log	31.96	96	0.3649	0.93	0.1057	0.8904	0.93	0.9307	0.0649	0.0112
No log	32.96	99	0.3597	0.935	0.1048	0.8894	0.935	0.9359	0.0613	0.0107
No log	33.96	102	0.3589	0.935	0.1056	0.8950	0.935	0.9359	0.0657	0.0106
No log	34.96	105	0.3595	0.94	0.1081	0.9086	0.94	0.9370	0.0614	0.0104
No log	35.96	108	0.3626	0.94	0.1100	0.9047	0.94	0.9370	0.0622	0.0098
No log	36.96	111	0.3520	0.94	0.1073	0.8958	0.94	0.9378	0.0579	0.0096
No log	37.96	114	0.3451	0.94	0.1033	0.8793	0.94	0.9378	0.0542	0.0094
No log	38.96	117	0.3427	0.935	0.1000	0.8685	0.935	0.9309	0.0604	0.0090
No log	39.96	120	0.3391	0.94	0.0977	0.8597	0.94	0.9353	0.0548	0.0089
No log	40.96	123	0.3363	0.95	0.0964	0.8537	0.9500	0.9522	0.0576	0.0088
No log	41.96	126	0.3458	0.95	0.1015	0.8524	0.9500	0.9522	0.0590	0.0087
No log	42.96	129	0.3618	0.935	0.1099	0.8628	0.935	0.9406	0.0640	0.0094
No log	43.96	132	0.3631	0.935	0.1109	0.8657	0.935	0.9406	0.0611	0.0093
No log	44.96	135	0.3622	0.94	0.1076	0.8571	0.94	0.9459	0.0555	0.0087
No log	45.96	138	0.3654	0.94	0.1066	0.8452	0.94	0.9459	0.0546	0.0083
No log	46.96	141	0.3672	0.935	0.1104	0.8382	0.935	0.9431	0.0588	0.0084
No log	47.96	144	0.3624	0.94	0.1081	0.8342	0.94	0.9482	0.0611	0.0085
No log	48.96	147	0.3603	0.945	0.1051	0.8251	0.945	0.9515	0.0590	0.0084
No log	49.96	150	0.3540	0.945	0.1013	0.8215	0.945	0.9515	0.0571	0.0082
No log	50.96	153	0.3538	0.945	0.0996	0.8169	0.945	0.9515	0.0600	0.0083
No log	51.96	156	0.3588	0.945	0.1014	0.8110	0.945	0.9515	0.0557	0.0082
No log	52.96	159	0.3600	0.945	0.1029	0.8046	0.945	0.9515	0.0597	0.0085
No log	53.96	162	0.3615	0.94	0.1037	0.7998	0.94	0.9398	0.0556	0.0083
No log	54.96	165	0.3589	0.94	0.1037	0.7987	0.94	0.9398	0.0562	0.0082
No log	55.96	168	0.3582	0.945	0.1037	0.7922	0.945	0.9515	0.0570	0.0083
No log	56.96	171	0.3695	0.945	0.1032	0.7887	0.945	0.9515	0.0590	0.0092
No log	57.96	174	0.3764	0.945	0.1008	0.7897	0.945	0.9490	0.0637	0.0098
No log	58.96	177	0.3719	0.955	0.0990	0.7909	0.955	0.9575	0.0616	0.0099
No log	59.96	180	0.3613	0.95	0.0977	0.7942	0.9500	0.9543	0.0550	0.0096
No log	60.96	183	0.3593	0.95	0.0984	0.7958	0.9500	0.9543	0.0573	0.0100
No log	61.96	186	0.3583	0.945	0.1005	0.7982	0.945	0.9512	0.0542	0.0102
No log	62.96	189	0.3594	0.94	0.1027	0.7994	0.94	0.9483	0.0558	0.0101
No log	63.96	192	0.3627	0.94	0.1042	0.7976	0.94	0.9483	0.0554	0.0101
No log	64.96	195	0.3668	0.94	0.1053	0.7944	0.94	0.9483	0.0560	0.0104
No log	65.96	198	0.3685	0.94	0.1063	0.7914	0.94	0.9483	0.0548	0.0104
No log	66.96	201	0.3701	0.94	0.1060	0.7880	0.94	0.9483	0.0563	0.0105
No log	67.96	204	0.3668	0.94	0.1047	0.7872	0.94	0.9483	0.0555	0.0111
No log	68.96	207	0.3682	0.945	0.1046	0.7863	0.945	0.9512	0.0587	0.0115
No log	69.96	210	0.3692	0.94	0.1054	0.7855	0.94	0.9425	0.0584	0.0117
No log	70.96	213	0.3734	0.94	0.1071	0.7829	0.94	0.9425	0.0582	0.0116
No log	71.96	216	0.3733	0.94	0.1079	0.7802	0.94	0.9425	0.0576	0.0116
No log	72.96	219	0.3720	0.935	0.1081	0.7769	0.935	0.9396	0.0579	0.0109
No log	73.96	222	0.3689	0.93	0.1074	0.7723	0.93	0.9296	0.0652	0.0096
No log	74.96	225	0.3687	0.935	0.1057	0.7701	0.935	0.9327	0.0634	0.0090
No log	75.96	228	0.3672	0.935	0.1053	0.7723	0.935	0.9327	0.0606	0.0083
No log	76.96	231	0.3649	0.935	0.1051	0.7772	0.935	0.9303	0.0575	0.0084
No log	77.96	234	0.3654	0.935	0.1055	0.7806	0.935	0.9303	0.0633	0.0084
No log	78.96	237	0.3630	0.935	0.1048	0.7823	0.935	0.9312	0.0647	0.0086
No log	79.96	240	0.3613	0.935	0.1036	0.7819	0.935	0.9312	0.0552	0.0088
No log	80.96	243	0.3578	0.94	0.1020	0.7808	0.94	0.9356	0.0550	0.0086
No log	81.96	246	0.3561	0.94	0.1009	0.7794	0.94	0.9356	0.0551	0.0086
No log	82.96	249	0.3573	0.935	0.1010	0.7767	0.935	0.9303	0.0557	0.0088
No log	83.96	252	0.3574	0.935	0.1010	0.7744	0.935	0.9303	0.0556	0.0089
No log	84.96	255	0.3574	0.935	0.1004	0.7696	0.935	0.9303	0.0562	0.0091
No log	85.96	258	0.3578	0.935	0.1003	0.7659	0.935	0.9303	0.0593	0.0088
No log	86.96	261	0.3569	0.935	0.0994	0.7640	0.935	0.9303	0.0599	0.0090
No log	87.96	264	0.3541	0.94	0.0984	0.7629	0.94	0.9390	0.0592	0.0094
No log	88.96	267	0.3529	0.94	0.0982	0.7613	0.94	0.9390	0.0555	0.0095
No log	89.96	270	0.3517	0.94	0.0984	0.7598	0.94	0.9390	0.0514	0.0095
No log	90.96	273	0.3528	0.94	0.0992	0.7586	0.94	0.9390	0.0516	0.0096
No log	91.96	276	0.3539	0.94	0.1004	0.7581	0.94	0.9390	0.0518	0.0095
No log	92.96	279	0.3551	0.94	0.1016	0.7583	0.94	0.9390	0.0559	0.0095
No log	93.96	282	0.3565	0.94	0.1028	0.7586	0.94	0.9390	0.0554	0.0093
No log	94.96	285	0.3572	0.94	0.1035	0.7588	0.94	0.9390	0.0554	0.0093
No log	95.96	288	0.3586	0.94	0.1041	0.7587	0.94	0.9390	0.0556	0.0092
No log	96.96	291	0.3586	0.94	0.1044	0.7586	0.94	0.9390	0.0607	0.0092
No log	97.96	294	0.3589	0.94	0.1047	0.7585	0.94	0.9390	0.0608	0.0092
No log	98.96	297	0.3590	0.935	0.1049	0.7584	0.935	0.9303	0.0608	0.0092
No log	99.96	300	0.3591	0.935	0.1049	0.7583	0.935	0.9303	0.0608	0.0092

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2