predict-perception-bert-focus-assassin

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: 0.2964
• Rmse: 0.8992
• Rmse Focus::a Sull'assassino: 0.8992
• Mae: 0.7331
• Mae Focus::a Sull'assassino: 0.7331
• R2: 0.6500
• R2 Focus::a Sull'assassino: 0.6500
• Cos: 0.7391
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.6131
• 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 Sull'assassino	Mae	Mae Focus::a Sull'assassino	R2	R2 Focus::a Sull'assassino	Cos	Pair	Rank	Neighbors	Rsa
1.0674	1.0	15	0.9851	1.6393	1.6393	1.5316	1.5316	-0.1633	-0.1633	0.1304	0.0	0.5	0.2457	nan
1.0099	2.0	30	0.8921	1.5601	1.5601	1.4317	1.4317	-0.0535	-0.0535	0.5652	0.0	0.5	0.4734	nan
0.9295	3.0	45	0.7345	1.4155	1.4155	1.3113	1.3113	0.1327	0.1327	0.5652	0.0	0.5	0.3596	nan
0.8485	4.0	60	0.7282	1.4094	1.4094	1.2678	1.2678	0.1401	0.1401	0.7391	0.0	0.5	0.5367	nan
0.7551	5.0	75	0.5966	1.2758	1.2758	1.1144	1.1144	0.2955	0.2955	0.6522	0.0	0.5	0.3911	nan
0.5563	6.0	90	0.4578	1.1175	1.1175	0.9105	0.9105	0.4594	0.4594	0.6522	0.0	0.5	0.3911	nan
0.4048	7.0	105	0.3539	0.9826	0.9826	0.7770	0.7770	0.5821	0.5821	0.6522	0.0	0.5	0.5522	nan
0.3319	8.0	120	0.2938	0.8953	0.8953	0.7110	0.7110	0.6530	0.6530	0.6522	0.0	0.5	0.6021	nan
0.2224	9.0	135	0.3455	0.9708	0.9708	0.7607	0.7607	0.5921	0.5921	0.6522	0.0	0.5	0.3911	nan
0.1794	10.0	150	0.2719	0.8612	0.8612	0.6768	0.6768	0.6790	0.6790	0.7391	0.0	0.5	0.6131	nan
0.1553	11.0	165	0.2855	0.8826	0.8826	0.7053	0.7053	0.6628	0.6628	0.7391	0.0	0.5	0.6131	nan
0.1008	12.0	180	0.3000	0.9046	0.9046	0.7255	0.7255	0.6458	0.6458	0.6522	0.0	0.5	0.5261	nan
0.1121	13.0	195	0.2817	0.8766	0.8766	0.7236	0.7236	0.6674	0.6674	0.7391	0.0	0.5	0.6131	nan
0.08	14.0	210	0.3504	0.9777	0.9777	0.7631	0.7631	0.5863	0.5863	0.7391	0.0	0.5	0.6131	nan
0.0802	15.0	225	0.3031	0.9094	0.9094	0.7565	0.7565	0.6420	0.6420	0.7391	0.0	0.5	0.6131	nan
0.0685	16.0	240	0.3041	0.9109	0.9109	0.7409	0.7409	0.6408	0.6408	0.7391	0.0	0.5	0.6131	nan
0.0592	17.0	255	0.3496	0.9767	0.9767	0.7812	0.7812	0.5871	0.5871	0.7391	0.0	0.5	0.6131	nan
0.0625	18.0	270	0.3260	0.9430	0.9430	0.7757	0.7757	0.6151	0.6151	0.7391	0.0	0.5	0.6131	nan
0.0589	19.0	285	0.3118	0.9222	0.9222	0.7442	0.7442	0.6318	0.6318	0.7391	0.0	0.5	0.6131	nan
0.0518	20.0	300	0.3062	0.9140	0.9140	0.7459	0.7459	0.6384	0.6384	0.7391	0.0	0.5	0.6131	nan
0.0456	21.0	315	0.3200	0.9344	0.9344	0.7592	0.7592	0.6221	0.6221	0.7391	0.0	0.5	0.6131	nan
0.0477	22.0	330	0.3132	0.9244	0.9244	0.7532	0.7532	0.6301	0.6301	0.7391	0.0	0.5	0.6131	nan
0.0448	23.0	345	0.3006	0.9056	0.9056	0.7321	0.7321	0.6450	0.6450	0.6522	0.0	0.5	0.5261	nan
0.0494	24.0	360	0.2985	0.9024	0.9024	0.7463	0.7463	0.6475	0.6475	0.7391	0.0	0.5	0.6131	nan
0.0369	25.0	375	0.3039	0.9105	0.9105	0.7359	0.7359	0.6412	0.6412	0.7391	0.0	0.5	0.6131	nan
0.0456	26.0	390	0.2989	0.9030	0.9030	0.7210	0.7210	0.6471	0.6471	0.7391	0.0	0.5	0.6131	nan
0.044	27.0	405	0.2997	0.9042	0.9042	0.7418	0.7418	0.6461	0.6461	0.7391	0.0	0.5	0.6131	nan
0.0352	28.0	420	0.2970	0.9001	0.9001	0.7346	0.7346	0.6493	0.6493	0.7391	0.0	0.5	0.6131	nan
0.0429	29.0	435	0.2970	0.9001	0.9001	0.7281	0.7281	0.6493	0.6493	0.7391	0.0	0.5	0.6131	nan
0.0378	30.0	450	0.2964	0.8992	0.8992	0.7331	0.7331	0.6500	0.6500	0.7391	0.0	0.5	0.6131	nan

Framework versions

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