Modelpublic

akseljoonas/biotech-sentiment-v3-A-deberta-v3-large

sourceHugging Facemitupdated 6mo agoView on Hugging Face

0likes6downloads

Model Card

biotech-sentiment-v3-A-deberta-v3-large

This model is a fine-tuned version of microsoft/deberta-v3-large on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.9956
—F1 Macro: 0.2401
—F1 Weighted: 0.4056
—Accuracy: 0.5630
—F1 Neg: 0.0
—F1 Neu: 0.7204
—F1 Pos: 0.0

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: 16
—evalbatchsize: 64
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 32
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 0.1
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Macro	F1 Weighted	Accuracy	F1 Neu	F1 Pos
102.3262	1.0	85	45.8907	0.1184	0.0767	0.2159	0.0	0.3551
17.9541	2.0	170	5.1985	0.1184	0.0767	0.2159	0.0	0.3551
3.4997	3.0	255	1.1731	0.2401	0.4056	0.5630	0.7204	0.0
2.4389	4.0	340	1.0647	0.2401	0.4056	0.5630	0.7204	0.0
2.6072	5.0	425	1.2260	0.2401	0.4056	0.5630	0.7204	0.0
2.3879	6.0	510	1.0551	0.2401	0.4056	0.5630	0.7204	0.0
2.6186	7.0	595	0.9950	0.2401	0.4056	0.5630	0.7204	0.0
2.1174	8.0	680	0.9923	0.2401	0.4056	0.5630	0.7204	0.0
2.0726	9.0	765	0.9892	0.2401	0.4056	0.5630	0.7204	0.0
2.1102	10.0	850	0.9956	0.2401	0.4056	0.5630	0.7204	0.0

Framework versions

—Transformers 5.3.0
—Pytorch 2.10.0+cu128
—Datasets 4.8.2
—Tokenizers 0.22.2