predict-perception-bert-cause-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: 1.6269
• Rmse: 1.2763
• Rmse Cause::a Spontanea, priva di un agente scatenante: 1.2763
• Mae: 1.0431
• Mae Cause::a Spontanea, priva di un agente scatenante: 1.0431
• R2: -1.4329
• R2 Cause::a Spontanea, priva di un agente scatenante: -1.4329
• Cos: -0.3913
• Pair: 0.0
• Rank: 0.5
• Neighbors: 0.3371
• 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 Spontanea, priva di un agente scatenante	Mae	Mae Cause::a Spontanea, priva di un agente scatenante	R2	R2 Cause::a Spontanea, priva di un agente scatenante	Cos	Pair	Rank	Neighbors	Rsa
0.994	1.0	15	0.7156	0.8465	0.8465	0.7809	0.7809	-0.0701	-0.0701	-0.1304	0.0	0.5	0.2971	nan
0.9757	2.0	30	0.7096	0.8429	0.8429	0.7666	0.7666	-0.0611	-0.0611	0.0435	0.0	0.5	0.2515	nan
1.0086	3.0	45	0.7779	0.8825	0.8825	0.7981	0.7981	-0.1632	-0.1632	-0.0435	0.0	0.5	0.2899	nan
0.9127	4.0	60	0.8158	0.9038	0.9038	0.8171	0.8171	-0.2199	-0.2199	-0.2174	0.0	0.5	0.2975	nan
0.8555	5.0	75	0.7691	0.8775	0.8775	0.8121	0.8121	-0.1501	-0.1501	-0.2174	0.0	0.5	0.3299	nan
0.8702	6.0	90	0.7818	0.8848	0.8848	0.7781	0.7781	-0.1691	-0.1691	0.0435	0.0	0.5	0.2515	nan
0.76	7.0	105	0.8377	0.9158	0.9158	0.7985	0.7985	-0.2526	-0.2526	0.0435	0.0	0.5	0.2515	nan
0.6997	8.0	120	0.9065	0.9527	0.9527	0.8370	0.8370	-0.3555	-0.3555	-0.2174	0.0	0.5	0.3147	nan
0.5963	9.0	135	1.0611	1.0308	1.0308	0.8396	0.8396	-0.5867	-0.5867	-0.0435	0.0	0.5	0.2645	nan
0.5413	10.0	150	1.1724	1.0835	1.0835	0.8649	0.8649	-0.7532	-0.7532	-0.0435	0.0	0.5	0.2645	nan
0.4994	11.0	165	1.1471	1.0717	1.0717	0.8857	0.8857	-0.7154	-0.7154	-0.2174	0.0	0.5	0.3271	nan
0.4208	12.0	180	1.2136	1.1024	1.1024	0.9392	0.9392	-0.8148	-0.8148	-0.2174	0.0	0.5	0.3169	nan
0.316	13.0	195	1.3499	1.1626	1.1626	0.9395	0.9395	-1.0187	-1.0187	-0.2174	0.0	0.5	0.3271	nan
0.2893	14.0	210	1.4229	1.1937	1.1937	0.9608	0.9608	-1.1278	-1.1278	-0.3043	0.0	0.5	0.3269	nan
0.235	15.0	225	1.4699	1.2132	1.2132	0.9785	0.9785	-1.1981	-1.1981	-0.0435	0.0	0.5	0.2865	nan
0.2397	16.0	240	1.5492	1.2455	1.2455	1.0005	1.0005	-1.3167	-1.3167	-0.0435	0.0	0.5	0.2655	nan
0.1973	17.0	255	1.5541	1.2474	1.2474	1.0165	1.0165	-1.3239	-1.3239	-0.0435	0.0	0.5	0.2655	nan
0.1793	18.0	270	1.4966	1.2242	1.2242	1.0058	1.0058	-1.2380	-1.2380	-0.3043	0.0	0.5	0.3437	nan
0.16	19.0	285	1.4977	1.2246	1.2246	1.0140	1.0140	-1.2396	-1.2396	-0.3913	0.0	0.5	0.3371	nan
0.1501	20.0	300	1.5751	1.2558	1.2558	1.0254	1.0254	-1.3553	-1.3553	-0.3913	0.0	0.5	0.3371	nan
0.1342	21.0	315	1.7011	1.3051	1.3051	1.0681	1.0681	-1.5438	-1.5438	-0.2174	0.0	0.5	0.2715	nan
0.137	22.0	330	1.5557	1.2481	1.2481	1.0393	1.0393	-1.3263	-1.3263	-0.3043	0.0	0.5	0.3437	nan
0.11	23.0	345	1.5475	1.2448	1.2448	1.0320	1.0320	-1.3141	-1.3141	-0.3913	0.0	0.5	0.3371	nan
0.1106	24.0	360	1.6006	1.2660	1.2660	1.0452	1.0452	-1.3936	-1.3936	-0.3913	0.0	0.5	0.3297	nan
0.1013	25.0	375	1.5907	1.2621	1.2621	1.0368	1.0368	-1.3787	-1.3787	-0.3043	0.0	0.5	0.2929	nan
0.0863	26.0	390	1.6436	1.2829	1.2829	1.0496	1.0496	-1.4578	-1.4578	-0.3043	0.0	0.5	0.2929	nan
0.0929	27.0	405	1.6000	1.2658	1.2658	1.0341	1.0341	-1.3927	-1.3927	-0.3043	0.0	0.5	0.3245	nan
0.0829	28.0	420	1.6277	1.2767	1.2767	1.0422	1.0422	-1.4341	-1.4341	-0.3913	0.0	0.5	0.3371	nan
0.0884	29.0	435	1.6324	1.2785	1.2785	1.0436	1.0436	-1.4411	-1.4411	-0.3913	0.0	0.5	0.3371	nan
0.0896	30.0	450	1.6269	1.2763	1.2763	1.0431	1.0431	-1.4329	-1.4329	-0.3913	0.0	0.5	0.3371	nan

Framework versions

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