Modelpublic

NasimB/gpt2-dp-mod-datasets-txt-processing-rarity-all

sourceHugging Facemitupdated 3y agoView on Hugging Face

0likes13downloads

Model Card

gpt2-dp-mod-datasets-txt-processing-rarity-all

This model is a fine-tuned version of gpt2 on the generator dataset. It achieves the following results on the evaluation set:

—Loss: 4.4242

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: 0.0005
—trainbatchsize: 64
—evalbatchsize: 64
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 1000
—num_epochs: 7
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
6.7606	0.29	500	5.6933
5.4375	0.59	1000	5.2559
5.0937	0.88	1500	5.0171
4.8204	1.18	2000	4.8701
4.6728	1.47	2500	4.7593
4.574	1.77	3000	4.6587
4.4456	2.06	3500	4.5885
4.258	2.36	4000	4.5468
4.2423	2.65	4500	4.4860
4.2036	2.94	5000	4.4302
3.9737	3.24	5500	4.4364
3.9439	3.53	6000	4.4019
3.9271	3.83	6500	4.3632
3.7901	4.12	7000	4.3689
3.6474	4.42	7500	4.3662
3.6414	4.71	8000	4.3472
3.6338	5.01	8500	4.3344
3.3764	5.3	9000	4.3618
3.3821	5.59	9500	4.3568
3.3777	5.89	10000	4.3513
3.2752	6.18	10500	4.3602
3.2228	6.48	11000	4.3652
3.2172	6.77	11500	4.3656

Framework versions

—Transformers 4.26.1
—Pytorch 1.11.0+cu113
—Datasets 2.13.0
—Tokenizers 0.13.3