predict-perception-bert-blame-concept

This model is a fine-tuned version of dbmdz/bert-base-italian-xxl-cased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7359
• Rmse: 0.6962
• Rmse Blame::a Un concetto astratto o un'emozione: 0.6962
• Mae: 0.5010
• Mae Blame::a Un concetto astratto o un'emozione: 0.5010
• R2: 0.3974
• R2 Blame::a Un concetto astratto o un'emozione: 0.3974
• Cos: 0.3913
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.5507
• Rsa: nan

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: 1e-05
• train_batch_size: 20
• eval_batch_size: 8
• seed: 1996
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Rmse	Rmse Blame::a Un concetto astratto o un'emozione	Mae	Mae Blame::a Un concetto astratto o un'emozione	R2	R2 Blame::a Un concetto astratto o un'emozione	Cos	Pair	Rank	Neighbors	Rsa
1.0979	1.0	15	1.2387	0.9033	0.9033	0.6603	0.6603	-0.0144	-0.0144	0.0435	0.0	0.5	0.3432	nan
1.0172	2.0	30	1.1498	0.8703	0.8703	0.5964	0.5964	0.0584	0.0584	0.0435	0.0	0.5	0.2935	nan
0.9879	3.0	45	1.2139	0.8942	0.8942	0.6197	0.6197	0.0060	0.0060	0.2174	0.0	0.5	0.4582	nan
0.9723	4.0	60	1.1152	0.8571	0.8571	0.5982	0.5982	0.0867	0.0867	0.2174	0.0	0.5	0.3921	nan
0.9584	5.0	75	1.0607	0.8358	0.8358	0.5959	0.5959	0.1314	0.1314	0.0435	0.0	0.5	0.4165	nan
0.9023	6.0	90	1.0031	0.8128	0.8128	0.5827	0.5827	0.1786	0.1786	-0.0435	0.0	0.5	0.3862	nan
0.8745	7.0	105	0.9715	0.7999	0.7999	0.5796	0.5796	0.2044	0.2044	0.3043	0.0	0.5	0.3665	nan
0.8082	8.0	120	0.8984	0.7692	0.7692	0.5699	0.5699	0.2643	0.2643	0.1304	0.0	0.5	0.3390	nan
0.7475	9.0	135	0.8532	0.7497	0.7497	0.5849	0.5849	0.3013	0.3013	0.0435	0.0	0.5	0.3100	nan
0.6599	10.0	150	0.8737	0.7586	0.7586	0.5822	0.5822	0.2846	0.2846	0.3043	0.0	0.5	0.3830	nan
0.5867	11.0	165	0.8159	0.7331	0.7331	0.5752	0.5752	0.3318	0.3318	0.2174	0.0	0.5	0.4439	nan
0.5081	12.0	180	0.8367	0.7424	0.7424	0.6071	0.6071	0.3148	0.3148	0.0435	0.0	0.5	0.3561	nan
0.4801	13.0	195	0.8353	0.7417	0.7417	0.5567	0.5567	0.3160	0.3160	0.3913	0.0	0.5	0.5850	nan
0.3714	14.0	210	0.8050	0.7282	0.7282	0.5824	0.5824	0.3408	0.3408	0.1304	0.0	0.5	0.3975	nan
0.3306	15.0	225	0.7833	0.7183	0.7183	0.5570	0.5570	0.3585	0.3585	0.2174	0.0	0.5	0.4604	nan
0.2674	16.0	240	0.8148	0.7326	0.7326	0.5475	0.5475	0.3328	0.3328	0.3043	0.0	0.5	0.4891	nan
0.2129	17.0	255	0.8715	0.7576	0.7576	0.5537	0.5537	0.2863	0.2863	0.4783	0.0	0.5	0.5017	nan
0.1924	18.0	270	0.7944	0.7234	0.7234	0.5276	0.5276	0.3495	0.3495	0.4783	0.0	0.5	0.5797	nan
0.1984	19.0	285	0.7885	0.7207	0.7207	0.5208	0.5208	0.3543	0.3543	0.3913	0.0	0.5	0.5507	nan
0.1623	20.0	300	0.7682	0.7113	0.7113	0.5132	0.5132	0.3709	0.3709	0.4783	0.0	0.5	0.5797	nan
0.1409	21.0	315	0.7653	0.7100	0.7100	0.5215	0.5215	0.3733	0.3733	0.3043	0.0	0.5	0.5415	nan
0.1386	22.0	330	0.7688	0.7116	0.7116	0.5124	0.5124	0.3704	0.3704	0.3913	0.0	0.5	0.5507	nan
0.123	23.0	345	0.7756	0.7148	0.7148	0.5144	0.5144	0.3648	0.3648	0.3913	0.0	0.5	0.5507	nan
0.1175	24.0	360	0.7423	0.6993	0.6993	0.5015	0.5015	0.3921	0.3921	0.3913	0.0	0.5	0.5507	nan
0.1188	25.0	375	0.7255	0.6913	0.6913	0.5063	0.5063	0.4059	0.4059	0.2174	0.0	0.5	0.4604	nan
0.1155	26.0	390	0.7635	0.7091	0.7091	0.5083	0.5083	0.3748	0.3748	0.4783	0.0	0.5	0.5797	nan
0.0981	27.0	405	0.7128	0.6852	0.6852	0.5020	0.5020	0.4163	0.4163	0.3043	0.0	0.5	0.5415	nan
0.1109	28.0	420	0.7430	0.6996	0.6996	0.5023	0.5023	0.3915	0.3915	0.3913	0.0	0.5	0.5507	nan
0.1081	29.0	435	0.7367	0.6966	0.6966	0.5007	0.5007	0.3967	0.3967	0.3913	0.0	0.5	0.5507	nan
0.0953	30.0	450	0.7359	0.6962	0.6962	0.5010	0.5010	0.3974	0.3974	0.3913	0.0	0.5	0.5507	nan

Framework versions

• Transformers 4.16.2
• Pytorch 1.10.2+cu113
• Datasets 1.18.3
• Tokenizers 0.11.0