predict-perception-bert-blame-victim

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: 0.5075
• Rmse: 0.4599
• Rmse Blame::a La vittima: 0.4599
• Mae: 0.3607
• Mae Blame::a La vittima: 0.3607
• R2: -0.1848
• R2 Blame::a La vittima: -0.1848
• Cos: 0.2174
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.2924
• 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 La vittima	Mae	Mae Blame::a La vittima	R2	R2 Blame::a La vittima	Cos	Pair	Rank	Neighbors	Rsa
1.0264	1.0	15	0.4334	0.4250	0.4250	0.3666	0.3666	-0.0119	-0.0119	0.1304	0.0	0.5	0.2703	nan
0.9814	2.0	30	0.4505	0.4333	0.4333	0.3744	0.3744	-0.0517	-0.0517	0.2174	0.0	0.5	0.2751	nan
0.9283	3.0	45	0.4349	0.4257	0.4257	0.3627	0.3627	-0.0152	-0.0152	0.1304	0.0	0.5	0.2779	nan
0.8904	4.0	60	0.4662	0.4408	0.4408	0.3773	0.3773	-0.0884	-0.0884	-0.0435	0.0	0.5	0.2681	nan
0.836	5.0	75	0.4188	0.4177	0.4177	0.3609	0.3609	0.0223	0.0223	0.2174	0.0	0.5	0.3051	nan
0.8293	6.0	90	0.4142	0.4155	0.4155	0.3512	0.3512	0.0330	0.0330	0.2174	0.0	0.5	0.3220	nan
0.7629	7.0	105	0.3837	0.3999	0.3999	0.3387	0.3387	0.1041	0.1041	0.2174	0.0	0.5	0.3051	nan
0.7266	8.0	120	0.3664	0.3907	0.3907	0.3250	0.3250	0.1446	0.1446	0.3043	0.0	0.5	0.3409	nan
0.6121	9.0	135	0.3718	0.3936	0.3936	0.3312	0.3312	0.1320	0.1320	0.3043	0.0	0.5	0.3983	nan
0.5694	10.0	150	0.3679	0.3915	0.3915	0.3197	0.3197	0.1411	0.1411	0.3913	0.0	0.5	0.3518	nan
0.4647	11.0	165	0.3868	0.4015	0.4015	0.3340	0.3340	0.0970	0.0970	0.2174	0.0	0.5	0.3285	nan
0.4212	12.0	180	0.3717	0.3936	0.3936	0.3188	0.3188	0.1322	0.1322	0.3913	0.0	0.5	0.3518	nan
0.3605	13.0	195	0.3437	0.3784	0.3784	0.3066	0.3066	0.1976	0.1976	0.3043	0.0	0.5	0.3423	nan
0.2759	14.0	210	0.3892	0.4027	0.4027	0.3230	0.3230	0.0914	0.0914	0.3913	0.0	0.5	0.3518	nan
0.2868	15.0	225	0.3720	0.3937	0.3937	0.3218	0.3218	0.1315	0.1315	0.3913	0.0	0.5	0.3440	nan
0.2467	16.0	240	0.3881	0.4022	0.4022	0.3291	0.3291	0.0939	0.0939	0.3043	0.0	0.5	0.3363	nan
0.2013	17.0	255	0.4121	0.4144	0.4144	0.3373	0.3373	0.0380	0.0380	0.3043	0.0	0.5	0.3363	nan
0.1966	18.0	270	0.4808	0.4476	0.4476	0.3506	0.3506	-0.1224	-0.1224	0.3913	0.0	0.5	0.3214	nan
0.177	19.0	285	0.4263	0.4215	0.4215	0.3398	0.3398	0.0046	0.0046	0.2174	0.0	0.5	0.2924	nan
0.1589	20.0	300	0.4274	0.4220	0.4220	0.3363	0.3363	0.0022	0.0022	0.2174	0.0	0.5	0.2924	nan
0.1488	21.0	315	0.4548	0.4353	0.4353	0.3431	0.3431	-0.0618	-0.0618	0.3043	0.0	0.5	0.2924	nan
0.1428	22.0	330	0.4405	0.4285	0.4285	0.3417	0.3417	-0.0285	-0.0285	0.3043	0.0	0.5	0.3363	nan
0.1294	23.0	345	0.4955	0.4544	0.4544	0.3565	0.3565	-0.1568	-0.1568	0.3913	0.0	0.5	0.3440	nan
0.1291	24.0	360	0.4861	0.4501	0.4501	0.3529	0.3529	-0.1348	-0.1348	0.2174	0.0	0.5	0.2924	nan
0.1187	25.0	375	0.4752	0.4450	0.4450	0.3518	0.3518	-0.1095	-0.1095	0.2174	0.0	0.5	0.2924	nan
0.1141	26.0	390	0.5131	0.4624	0.4624	0.3598	0.3598	-0.1978	-0.1978	0.3043	0.0	0.5	0.2924	nan
0.1094	27.0	405	0.4863	0.4502	0.4502	0.3547	0.3547	-0.1353	-0.1353	0.2174	0.0	0.5	0.2924	nan
0.0925	28.0	420	0.4900	0.4519	0.4519	0.3564	0.3564	-0.1439	-0.1439	0.2174	0.0	0.5	0.2924	nan
0.108	29.0	435	0.5019	0.4573	0.4573	0.3590	0.3590	-0.1719	-0.1719	0.2174	0.0	0.5	0.2924	nan
0.1054	30.0	450	0.5075	0.4599	0.4599	0.3607	0.3607	-0.1848	-0.1848	0.2174	0.0	0.5	0.2924	nan

Framework versions

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