Modelpublic

thrunlab/t5-large_boolq_dense_epochs-5

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes9downloads

Model Card

t5-largeboolqdense_epochs-5

This model is a fine-tuned version of t5-large on the super_glue dataset. It achieves the following results on the evaluation set:

—Loss: 0.3715
—Accuracy: 0.8462

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: 8
—evalbatchsize: 16
—seed: 0
—gradientaccumulationsteps: 4
—totaltrainbatch_size: 32
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 20
—num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6792	0.17	50	0.6652	0.6217
0.66	0.34	100	0.6595	0.6220
0.6614	0.51	150	0.6548	0.6232
0.636	0.68	200	0.6122	0.6985
0.4882	0.85	250	0.4702	0.7847
0.5068	1.02	300	0.4639	0.7862
0.3332	1.19	350	0.5297	0.7908
0.4296	1.36	400	0.3955	0.8373
0.356	1.53	450	0.4013	0.8410
0.3227	1.7	500	0.3715	0.8462
0.3516	1.87	550	0.3724	0.8428
0.2169	2.04	600	0.3906	0.8477
0.2199	2.21	650	0.4061	0.8572
0.1969	2.37	700	0.4351	0.8550
0.2713	2.54	750	0.5411	0.8584
0.2458	2.71	800	0.3924	0.8627
0.2134	2.88	850	0.3973	0.8630
0.1636	3.05	900	0.4933	0.8590
0.1108	3.22	950	0.9926	0.8621
0.1433	3.39	1000	0.6679	0.8602

Framework versions

—Transformers 4.34.1
—Pytorch 2.1.0+cu118
—Datasets 2.14.6
—Tokenizers 0.14.1