Modelpublic

wiorz/bert_small_summarized

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes8downloads

Model Card

bertsmallsummarized

This model is a fine-tuned version of bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

—Loss: 2.1652
—Accuracy: 0.82
—Precision: 0.4667
—Recall: 0.2
—F1: 0.2800
—D-index: 1.5200

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: 5e-05
—trainbatchsize: 4
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 1600
—num_epochs: 20
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	D-index
No log	1.0	200	0.4533	0.825	0.0	0.0	0.0	1.4529
No log	2.0	400	0.4694	0.825	0.0	0.0	0.0	1.4529
0.5094	3.0	600	0.6237	0.825	0.0	0.0	0.0	1.4529
0.5094	4.0	800	0.7898	0.81	0.4286	0.2571	0.3214	1.5270
0.3984	5.0	1000	0.9268	0.83	0.5556	0.1429	0.2273	1.5127
0.3984	6.0	1200	1.3541	0.8	0.4074	0.3143	0.3548	1.5339
0.3984	7.0	1400	1.4264	0.805	0.375	0.1714	0.2353	1.4893
0.0939	8.0	1600	1.8870	0.8	0.4194	0.3714	0.3939	1.5539
0.0939	9.0	1800	1.8734	0.825	0.5	0.1143	0.1860	1.4955
0.0061	10.0	2000	1.8938	0.825	0.5	0.1714	0.2553	1.5164
0.0061	11.0	2200	2.0755	0.825	0.5	0.1143	0.1860	1.4955
0.0061	12.0	2400	2.1068	0.805	0.4231	0.3143	0.3607	1.5406
0.0134	13.0	2600	2.0895	0.82	0.4444	0.1143	0.1818	1.4887
0.0134	14.0	2800	2.0520	0.815	0.4545	0.2857	0.3509	1.5439
0.0011	15.0	3000	2.0795	0.81	0.4211	0.2286	0.2963	1.5168
0.0011	16.0	3200	2.1177	0.815	0.4444	0.2286	0.3019	1.5235
0.0011	17.0	3400	2.1396	0.815	0.4444	0.2286	0.3019	1.5235
0.0003	18.0	3600	2.1605	0.825	0.5	0.2286	0.3137	1.5370
0.0003	19.0	3800	2.1677	0.825	0.5	0.2286	0.3137	1.5370
0.0	20.0	4000	2.1652	0.82	0.4667	0.2	0.2800	1.5200

Framework versions

—Transformers 4.28.0
—Pytorch 2.0.1+cu118
—Datasets 2.12.0
—Tokenizers 0.13.3