Modelpublic

avsolatorio/doc-topic-model_eval-02_train-04

sourceHugging Facemitupdated 2y agoView on Hugging Face

0likes8downloads

Model Card

doc-topic-modeleval-02train-04

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

—Loss: 0.0381
—Accuracy: 0.9879
—F1: 0.6381
—Precision: 0.7172
—Recall: 0.5747

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: 4
—evalbatchsize: 256
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 100
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.0934	0.4929	1000	0.0904	0.9814	0.0	0.0	0.0
0.0778	0.9857	2000	0.0702	0.9814	0.0	0.0	0.0
0.0618	1.4786	3000	0.0567	0.9828	0.1751	0.8200	0.0980
0.0535	1.9714	4000	0.0492	0.9843	0.3358	0.7842	0.2136
0.0473	2.4643	5000	0.0455	0.9855	0.4668	0.7407	0.3408
0.0436	2.9571	6000	0.0426	0.9861	0.5016	0.7524	0.3763
0.0389	3.4500	7000	0.0407	0.9865	0.5337	0.7497	0.4143
0.0376	3.9428	8000	0.0399	0.9866	0.5634	0.7187	0.4633
0.0339	4.4357	9000	0.0389	0.9870	0.5653	0.7472	0.4547
0.0337	4.9285	10000	0.0385	0.9873	0.5815	0.7552	0.4728
0.0295	5.4214	11000	0.0377	0.9872	0.6024	0.7156	0.5202
0.0305	5.9142	12000	0.0383	0.9874	0.5992	0.7317	0.5074
0.0254	6.4071	13000	0.0375	0.9876	0.6141	0.7281	0.5310
0.0273	6.9000	14000	0.0379	0.9877	0.6163	0.7325	0.5319
0.0228	7.3928	15000	0.0379	0.9877	0.6165	0.7367	0.5300
0.0235	7.8857	16000	0.0379	0.9874	0.6298	0.6930	0.5773
0.0208	8.3785	17000	0.0379	0.9877	0.6341	0.7129	0.5710
0.0204	8.8714	18000	0.0381	0.9879	0.6381	0.7172	0.5747

Framework versions

—Transformers 4.44.2
—Pytorch 2.4.1+cu121
—Datasets 2.21.0
—Tokenizers 0.19.1