Modelpublic

lupobricco/prosat_finetuned_pro_emit_correlations1

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes6downloads

Model Card

prosatfinetunedproemitcorrelations1

This model is a fine-tuned version of MilaNLProc/feel-it-italian-emotion on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.3056
—F1: 0.5497
—Roc Auc: 0.7451
—Accuracy: 0.3814

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: 2e-05
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.2939	1.0	1037	0.2836	0.3052	0.5938	0.2543
0.2378	2.0	2074	0.2466	0.4341	0.6536	0.3196
0.1988	3.0	3111	0.2483	0.5111	0.7094	0.3522
0.1681	4.0	4148	0.2487	0.5059	0.7031	0.3849
0.1431	5.0	5185	0.2599	0.5252	0.7201	0.3780
0.1213	6.0	6222	0.2681	0.5266	0.7243	0.3729
0.107	7.0	7259	0.2913	0.5335	0.7364	0.3729
0.092	8.0	8296	0.2920	0.5385	0.7337	0.3746
0.0857	9.0	9333	0.3048	0.5488	0.7464	0.3746
0.0753	10.0	10370	0.3056	0.5497	0.7451	0.3814

Framework versions

—Transformers 4.40.1
—Pytorch 2.3.0+cu118
—Datasets 2.19.0
—Tokenizers 0.19.1