Modelpublic

henryscheible/xlnet-base-cased_stereoset_finetuned

sourceHugging Facemitupdated 4y agoView on Hugging Face

0likes37downloads

Model Card

xlnet-base-casedstereosetfinetuned

This model is a fine-tuned version of xlnet-base-cased on the stereoset dataset. It achieves the following results on the evaluation set:

—Loss: 1.0332
—Accuracy: 0.7441

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: 128
—evalbatchsize: 64
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.21	5	0.7165	0.5055
No log	0.42	10	0.6932	0.5
No log	0.62	15	0.6971	0.5047
No log	0.83	20	0.7107	0.4953
No log	1.04	25	0.6895	0.5047
No log	1.25	30	0.6715	0.5840
No log	1.46	35	0.6476	0.6476
No log	1.67	40	0.6150	0.6970
No log	1.88	45	0.6170	0.6884
No log	2.08	50	0.6065	0.6797
No log	2.29	55	0.5865	0.7033
No log	2.5	60	0.5899	0.7064
No log	2.71	65	0.5980	0.7151
No log	2.92	70	0.5890	0.7229
No log	3.12	75	0.5930	0.7190
No log	3.33	80	0.6430	0.7049
No log	3.54	85	0.6677	0.7198
No log	3.75	90	0.6076	0.7370
No log	3.96	95	0.6041	0.7339
No log	4.17	100	0.6324	0.7323
No log	4.38	105	0.6990	0.7308
No log	4.58	110	0.7081	0.7433
No log	4.79	115	0.6549	0.7237
No log	5.0	120	0.6868	0.7072
No log	5.21	125	0.6525	0.7363
No log	5.42	130	0.7622	0.7418
No log	5.62	135	0.7730	0.7402
No log	5.83	140	0.7788	0.7449
No log	6.04	145	0.7609	0.7347
No log	6.25	150	0.8058	0.7323
No log	6.46	155	0.8525	0.7331
No log	6.67	160	0.8504	0.7339
No log	6.88	165	0.8424	0.7300
No log	7.08	170	0.8413	0.7394
No log	7.29	175	0.8808	0.7268
No log	7.5	180	0.9058	0.7292
No log	7.71	185	0.9338	0.7363
No log	7.92	190	0.9412	0.7370
No log	8.12	195	0.9453	0.7339
No log	8.33	200	0.9544	0.7394
No log	8.54	205	0.9664	0.7402
No log	8.75	210	0.9840	0.7339
No log	8.96	215	0.9896	0.7370
No log	9.17	220	1.0239	0.7410
No log	9.38	225	1.0306	0.7418
No log	9.58	230	1.0358	0.7402
No log	9.79	235	1.0351	0.7410
No log	10.0	240	1.0332	0.7441

Framework versions

—Transformers 4.26.1
—Pytorch 1.13.1
—Datasets 2.9.0
—Tokenizers 0.13.2