predict-perception-bert-blame-none

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: 0.8646
• Rmse: 1.1072
• Rmse Blame::a Nessuno: 1.1072
• Mae: 0.8721
• Mae Blame::a Nessuno: 0.8721
• R2: 0.3083
• R2 Blame::a Nessuno: 0.3083
• Cos: 0.5652
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.5070
• 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 Nessuno	Mae	Mae Blame::a Nessuno	R2	R2 Blame::a Nessuno	Cos	Pair	Rank	Neighbors	Rsa
1.007	1.0	15	1.2585	1.3358	1.3358	1.1752	1.1752	-0.0068	-0.0068	-0.0435	0.0	0.5	0.2970	nan
0.927	2.0	30	1.1310	1.2663	1.2663	1.0633	1.0633	0.0952	0.0952	0.4783	0.0	0.5	0.4012	nan
0.8376	3.0	45	1.0603	1.2261	1.2261	1.0574	1.0574	0.1518	0.1518	0.1304	0.0	0.5	0.2970	nan
0.7154	4.0	60	0.8347	1.0879	1.0879	0.8854	0.8854	0.3323	0.3323	0.6522	0.0	0.5	0.5209	nan
0.5766	5.0	75	0.7426	1.0261	1.0261	0.8340	0.8340	0.4059	0.4059	0.6522	0.0	0.5	0.5209	nan
0.4632	6.0	90	0.6671	0.9725	0.9725	0.7932	0.7932	0.4663	0.4663	0.6522	0.0	0.5	0.5209	nan
0.3854	7.0	105	0.6447	0.9561	0.9561	0.7424	0.7424	0.4842	0.4842	0.6522	0.0	0.5	0.4307	nan
0.3154	8.0	120	0.7198	1.0102	1.0102	0.8113	0.8113	0.4241	0.4241	0.6522	0.0	0.5	0.4307	nan
0.2637	9.0	135	0.7221	1.0118	1.0118	0.8319	0.8319	0.4223	0.4223	0.5652	0.0	0.5	0.4150	nan
0.1962	10.0	150	0.6999	0.9962	0.9962	0.7945	0.7945	0.4401	0.4401	0.4783	0.0	0.5	0.4056	nan
0.1784	11.0	165	0.7335	1.0198	1.0198	0.7969	0.7969	0.4132	0.4132	0.5652	0.0	0.5	0.4150	nan
0.1531	12.0	180	0.8277	1.0833	1.0833	0.8839	0.8839	0.3378	0.3378	0.4783	0.0	0.5	0.4440	nan
0.1425	13.0	195	0.8644	1.1070	1.1070	0.8726	0.8726	0.3085	0.3085	0.5652	0.0	0.5	0.5070	nan
0.0921	14.0	210	0.8874	1.1217	1.1217	0.9024	0.9024	0.2900	0.2900	0.4783	0.0	0.5	0.4440	nan
0.0913	15.0	225	0.8663	1.1083	1.1083	0.8914	0.8914	0.3070	0.3070	0.5652	0.0	0.5	0.5070	nan
0.08	16.0	240	0.8678	1.1093	1.1093	0.8762	0.8762	0.3057	0.3057	0.6522	0.0	0.5	0.5931	nan
0.0725	17.0	255	0.8497	1.0976	1.0976	0.8868	0.8868	0.3202	0.3202	0.4783	0.0	0.5	0.4440	nan
0.0696	18.0	270	0.8533	1.1000	1.1000	0.8796	0.8796	0.3173	0.3173	0.5652	0.0	0.5	0.5070	nan
0.0632	19.0	285	0.8563	1.1018	1.1018	0.8768	0.8768	0.3150	0.3150	0.5652	0.0	0.5	0.5070	nan
0.0511	20.0	300	0.8433	1.0935	1.0935	0.8684	0.8684	0.3254	0.3254	0.5652	0.0	0.5	0.5070	nan
0.0517	21.0	315	0.8449	1.0945	1.0945	0.8758	0.8758	0.3240	0.3240	0.4783	0.0	0.5	0.4440	nan
0.0556	22.0	330	0.8305	1.0851	1.0851	0.8469	0.8469	0.3356	0.3356	0.5652	0.0	0.5	0.5070	nan
0.0457	23.0	345	0.8369	1.0893	1.0893	0.8555	0.8555	0.3305	0.3305	0.5652	0.0	0.5	0.5070	nan
0.0496	24.0	360	0.8441	1.0940	1.0940	0.8648	0.8648	0.3247	0.3247	0.5652	0.0	0.5	0.5070	nan
0.0467	25.0	375	0.8470	1.0959	1.0959	0.8633	0.8633	0.3224	0.3224	0.5652	0.0	0.5	0.5070	nan
0.0446	26.0	390	0.8562	1.1018	1.1018	0.8708	0.8708	0.3151	0.3151	0.4783	0.0	0.5	0.4440	nan
0.0476	27.0	405	0.8600	1.1042	1.1042	0.8714	0.8714	0.3120	0.3120	0.5652	0.0	0.5	0.5070	nan
0.042	28.0	420	0.8657	1.1079	1.1079	0.8763	0.8763	0.3074	0.3074	0.4783	0.0	0.5	0.4440	nan
0.0431	29.0	435	0.8654	1.1077	1.1077	0.8734	0.8734	0.3077	0.3077	0.5652	0.0	0.5	0.5070	nan
0.0423	30.0	450	0.8646	1.1072	1.1072	0.8721	0.8721	0.3083	0.3083	0.5652	0.0	0.5	0.5070	nan

Framework versions

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