mobilebert_sa_GLUE_Experiment_logit_kd_data_aug_stsb_128

This model is a fine-tuned version of google/mobilebert-uncased on the GLUE STSB dataset. It achieves the following results on the evaluation set:

• Loss: 1.4602
• Pearson: 0.1596
• Spearmanr: 0.1582
• Combined Score: 0.1589

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
• train_batch_size: 128
• eval_batch_size: 128
• seed: 10
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Pearson	Spearmanr	Combined Score
0.5444	1.0	2518	1.4965	0.1589	0.1763	0.1676
0.3254	2.0	5036	1.5276	0.1502	0.1674	0.1588
0.2847	3.0	7554	1.5430	0.1587	0.1680	0.1634
0.2376	4.0	10072	1.6906	0.1669	0.1786	0.1728
0.1741	5.0	12590	1.4788	0.1662	0.1725	0.1694
0.1315	6.0	15108	1.5662	0.1640	0.1700	0.1670
0.1055	7.0	17626	1.5100	0.1663	0.1698	0.1680
0.0879	8.0	20144	1.4602	0.1596	0.1582	0.1589
0.0739	9.0	22662	1.6612	0.1584	0.1621	0.1603
0.0632	10.0	25180	1.5825	0.1489	0.1547	0.1518
0.0548	11.0	27698	1.5946	0.1421	0.1461	0.1441
0.0473	12.0	30216	1.6515	0.1526	0.1548	0.1537
0.0415	13.0	32734	1.6544	0.1506	0.1478	0.1492

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.9.0
• Tokenizers 0.13.2