predict-perception-bert-cause-object

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: 0.4120
• Rmse: 1.0345
• Rmse Cause::a Causata da un oggetto (es. una pistola): 1.0345
• Mae: 0.6181
• Mae Cause::a Causata da un oggetto (es. una pistola): 0.6181
• R2: 0.3837
• R2 Cause::a Causata da un oggetto (es. una pistola): 0.3837
• Cos: 0.9130
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.8986
• 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 Cause::a Causata da un oggetto (es. una pistola)	Mae	Mae Cause::a Causata da un oggetto (es. una pistola)	R2	R2 Cause::a Causata da un oggetto (es. una pistola)	Cos	Pair	Rank	Neighbors	Rsa
1.0824	1.0	15	0.6651	1.3143	1.3143	1.0930	1.0930	0.0052	0.0052	0.3043	0.0	0.5	0.4393	nan
0.9574	2.0	30	0.7088	1.3568	1.3568	1.1945	1.1945	-0.0601	-0.0601	0.0435	0.0	0.5	0.3380	nan
0.8151	3.0	45	0.6300	1.2791	1.2791	1.0206	1.0206	0.0577	0.0577	0.3043	0.0	0.5	0.3613	nan
0.6401	4.0	60	0.4871	1.1247	1.1247	0.7285	0.7285	0.2715	0.2715	0.5652	0.0	0.5	0.6424	nan
0.448	5.0	75	0.5005	1.1401	1.1401	0.7216	0.7216	0.2514	0.2514	0.4783	0.0	0.5	0.6077	nan
0.2893	6.0	90	0.4761	1.1119	1.1119	0.7237	0.7237	0.2879	0.2879	0.5652	0.0	0.5	0.6348	nan
0.174	7.0	105	0.4771	1.1131	1.1131	0.6836	0.6836	0.2865	0.2865	0.6522	0.0	0.5	0.6785	nan
0.1383	8.0	120	0.4313	1.0583	1.0583	0.6462	0.6462	0.3550	0.3550	0.8261	0.0	0.5	0.7586	nan
0.1105	9.0	135	0.4660	1.1001	1.1001	0.6737	0.6737	0.3030	0.3030	0.8261	0.0	0.5	0.7586	nan
0.0903	10.0	150	0.4866	1.1241	1.1241	0.7192	0.7192	0.2723	0.2723	0.7391	0.0	0.5	0.6833	nan
0.0571	11.0	165	0.4361	1.0642	1.0642	0.6130	0.6130	0.3478	0.3478	0.8261	0.0	0.5	0.7586	nan
0.0623	12.0	180	0.4578	1.0904	1.0904	0.6844	0.6844	0.3152	0.3152	0.6522	0.0	0.5	0.6785	nan
0.0526	13.0	195	0.4605	1.0936	1.0936	0.6697	0.6697	0.3112	0.3112	0.6522	0.0	0.5	0.6785	nan
0.0472	14.0	210	0.4440	1.0738	1.0738	0.6589	0.6589	0.3360	0.3360	0.7391	0.0	0.5	0.7327	nan
0.0492	15.0	225	0.4593	1.0922	1.0922	0.6812	0.6812	0.3130	0.3130	0.7391	0.0	0.5	0.6833	nan
0.0389	16.0	240	0.4195	1.0437	1.0437	0.6252	0.6252	0.3726	0.3726	0.8261	0.0	0.5	0.7586	nan
0.0396	17.0	255	0.4087	1.0302	1.0302	0.6119	0.6119	0.3888	0.3888	0.9130	0.0	0.5	0.8986	nan
0.0328	18.0	270	0.4274	1.0535	1.0535	0.6457	0.6457	0.3608	0.3608	0.8261	0.0	0.5	0.7431	nan
0.0345	19.0	285	0.4306	1.0574	1.0574	0.6576	0.6576	0.3560	0.3560	0.8261	0.0	0.5	0.7431	nan
0.0328	20.0	300	0.4067	1.0277	1.0277	0.6160	0.6160	0.3918	0.3918	0.9130	0.0	0.5	0.8986	nan
0.0344	21.0	315	0.4056	1.0263	1.0263	0.5948	0.5948	0.3934	0.3934	0.9130	0.0	0.5	0.8986	nan
0.0312	22.0	330	0.4236	1.0488	1.0488	0.6277	0.6277	0.3665	0.3665	0.9130	0.0	0.5	0.8986	nan
0.0241	23.0	345	0.4272	1.0533	1.0533	0.6444	0.6444	0.3610	0.3610	0.8261	0.0	0.5	0.7431	nan
0.0302	24.0	360	0.4046	1.0250	1.0250	0.6030	0.6030	0.3949	0.3949	0.8261	0.0	0.5	0.7586	nan
0.0244	25.0	375	0.4194	1.0436	1.0436	0.6320	0.6320	0.3728	0.3728	0.9130	0.0	0.5	0.8986	nan
0.0259	26.0	390	0.4025	1.0224	1.0224	0.6009	0.6009	0.3980	0.3980	0.8261	0.0	0.5	0.7586	nan
0.0265	27.0	405	0.4103	1.0323	1.0323	0.6180	0.6180	0.3863	0.3863	0.9130	0.0	0.5	0.8986	nan
0.0184	28.0	420	0.4059	1.0268	1.0268	0.6046	0.6046	0.3929	0.3929	0.8261	0.0	0.5	0.7586	nan
0.0257	29.0	435	0.4088	1.0304	1.0304	0.6122	0.6122	0.3885	0.3885	0.9130	0.0	0.5	0.8986	nan
0.0262	30.0	450	0.4120	1.0345	1.0345	0.6181	0.6181	0.3837	0.3837	0.9130	0.0	0.5	0.8986	nan

Framework versions

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