Modelpublic

NasimB/children_bnc_rarity_all_no_cut

sourceHugging Facemitupdated 3y agoView on Hugging Face

0likes23downloads

Model Card

childrenbncrarityallno_cut

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

—Loss: 4.3266

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: 6
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
6.7047	0.29	500	5.6398
5.3501	0.58	1000	5.2066
5.0056	0.88	1500	4.9588
4.7258	1.17	2000	4.8173
4.5734	1.46	2500	4.6948
4.4663	1.75	3000	4.5804
4.3402	2.05	3500	4.5071
4.1471	2.34	4000	4.4576
4.1137	2.63	4500	4.4027
4.0777	2.92	5000	4.3468
3.8629	3.22	5500	4.3449
3.8078	3.51	6000	4.3108
3.8044	3.8	6500	4.2763
3.7029	4.09	7000	4.2803
3.5324	4.39	7500	4.2741
3.5239	4.68	8000	4.2585
3.5091	4.97	8500	4.2454
3.3521	5.26	9000	4.2592
3.3357	5.56	9500	4.2584
3.3348	5.85	10000	4.2573

Framework versions

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