predict-perception-bert-focus-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: 0.8129
• Rmse: 1.0197
• Rmse Focus::a Su un concetto astratto o un'emozione: 1.0197
• Mae: 0.7494
• Mae Focus::a Su un concetto astratto o un'emozione: 0.7494
• R2: 0.1970
• R2 Focus::a Su un concetto astratto o un'emozione: 0.1970
• Cos: 0.4783
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.4667
• 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 Focus::a Su un concetto astratto o un'emozione	Mae	Mae Focus::a Su un concetto astratto o un'emozione	R2	R2 Focus::a Su un concetto astratto o un'emozione	Cos	Pair	Rank	Neighbors	Rsa
1.047	1.0	15	1.0199	1.1422	1.1422	0.9321	0.9321	-0.0075	-0.0075	0.1304	0.0	0.5	0.3199	nan
0.9914	2.0	30	0.9724	1.1153	1.1153	0.9407	0.9407	0.0393	0.0393	0.2174	0.0	0.5	0.3954	nan
0.9049	3.0	45	0.9406	1.0969	1.0969	0.9170	0.9170	0.0708	0.0708	0.2174	0.0	0.5	0.3632	nan
0.8826	4.0	60	0.8553	1.0460	1.0460	0.8570	0.8570	0.1551	0.1551	0.2174	0.0	0.5	0.3230	nan
0.7837	5.0	75	0.8324	1.0319	1.0319	0.8683	0.8683	0.1776	0.1776	0.2174	0.0	0.5	0.3419	nan
0.7013	6.0	90	0.7737	0.9949	0.9949	0.8150	0.8150	0.2356	0.2356	0.5652	0.0	0.5	0.5023	nan
0.6429	7.0	105	0.7832	1.0010	1.0010	0.8005	0.8005	0.2262	0.2262	0.3913	0.0	0.5	0.4446	nan
0.5526	8.0	120	0.7734	0.9946	0.9946	0.7704	0.7704	0.2360	0.2360	0.3043	0.0	0.5	0.2923	nan
0.5194	9.0	135	0.6624	0.9205	0.9205	0.7013	0.7013	0.3456	0.3456	0.3913	0.0	0.5	0.3523	nan
0.4278	10.0	150	0.8255	1.0276	1.0276	0.7351	0.7351	0.1845	0.1845	0.3043	0.0	0.5	0.4349	nan
0.3522	11.0	165	0.9340	1.0931	1.0931	0.8069	0.8069	0.0773	0.0773	0.3913	0.0	0.5	0.4059	nan
0.314	12.0	180	0.7495	0.9792	0.9792	0.7254	0.7254	0.2596	0.2596	0.3913	0.0	0.5	0.4059	nan
0.2665	13.0	195	0.8574	1.0473	1.0473	0.7678	0.7678	0.1530	0.1530	0.3913	0.0	0.5	0.4059	nan
0.2348	14.0	210	0.7913	1.0061	1.0061	0.7218	0.7218	0.2183	0.2183	0.3913	0.0	0.5	0.4059	nan
0.1859	15.0	225	0.8012	1.0124	1.0124	0.7162	0.7162	0.2085	0.2085	0.3913	0.0	0.5	0.4059	nan
0.1373	16.0	240	0.8405	1.0369	1.0369	0.7318	0.7318	0.1697	0.1697	0.3043	0.0	0.5	0.3734	nan
0.1245	17.0	255	0.8398	1.0365	1.0365	0.7455	0.7455	0.1703	0.1703	0.4783	0.0	0.5	0.4667	nan
0.1148	18.0	270	0.7948	1.0083	1.0083	0.7140	0.7140	0.2148	0.2148	0.3913	0.0	0.5	0.4175	nan
0.1187	19.0	285	0.8301	1.0305	1.0305	0.7381	0.7381	0.1799	0.1799	0.3913	0.0	0.5	0.4175	nan
0.1236	20.0	300	0.8867	1.0650	1.0650	0.7879	0.7879	0.1240	0.1240	0.3913	0.0	0.5	0.4059	nan
0.1101	21.0	315	0.8405	1.0369	1.0369	0.7632	0.7632	0.1696	0.1696	0.3913	0.0	0.5	0.4059	nan
0.0902	22.0	330	0.7850	1.0021	1.0021	0.7173	0.7173	0.2245	0.2245	0.3043	0.0	0.5	0.3734	nan
0.093	23.0	345	0.7386	0.9720	0.9720	0.6960	0.6960	0.2704	0.2704	0.3913	0.0	0.5	0.4175	nan
0.0846	24.0	360	0.7748	0.9956	0.9956	0.7150	0.7150	0.2345	0.2345	0.3913	0.0	0.5	0.4175	nan
0.0826	25.0	375	0.7951	1.0085	1.0085	0.7230	0.7230	0.2145	0.2145	0.3913	0.0	0.5	0.4175	nan
0.0749	26.0	390	0.8470	1.0409	1.0409	0.7621	0.7621	0.1633	0.1633	0.4783	0.0	0.5	0.4667	nan
0.069	27.0	405	0.7968	1.0096	1.0096	0.7275	0.7275	0.2129	0.2129	0.3913	0.0	0.5	0.4175	nan
0.0775	28.0	420	0.8298	1.0303	1.0303	0.7589	0.7589	0.1802	0.1802	0.4783	0.0	0.5	0.4667	nan
0.0783	29.0	435	0.8113	1.0188	1.0188	0.7469	0.7469	0.1985	0.1985	0.4783	0.0	0.5	0.4667	nan
0.0773	30.0	450	0.8129	1.0197	1.0197	0.7494	0.7494	0.1970	0.1970	0.4783	0.0	0.5	0.4667	nan

Framework versions

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