distilbert_base_lda_stsb

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

• Loss: 0.9069
• Pearson: 0.7831
• Spearmanr: 0.7812
• Combined Score: 0.7822

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: 256
• eval_batch_size: 256
• seed: 10
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Pearson	Spearmanr	Combined Score
2.63	1.0	23	2.2703	0.2049	0.1869	0.1959
1.6161	2.0	46	1.5326	0.5968	0.5988	0.5978
1.0577	3.0	69	0.9897	0.7544	0.7530	0.7537
0.7651	4.0	92	1.3080	0.7422	0.7513	0.7468
0.5755	5.0	115	1.0189	0.7718	0.7729	0.7724
0.4879	6.0	138	0.9542	0.7654	0.7639	0.7647
0.3923	7.0	161	0.9830	0.7704	0.7702	0.7703
0.2951	8.0	184	1.0897	0.7662	0.7689	0.7675
0.2586	9.0	207	1.1433	0.7597	0.7595	0.7596
0.2193	10.0	230	0.9188	0.7807	0.7783	0.7795
0.2001	11.0	253	0.9884	0.7733	0.7733	0.7733
0.1817	12.0	276	1.0962	0.7711	0.7708	0.7710
0.1645	13.0	299	0.9747	0.7704	0.7695	0.7700
0.1677	14.0	322	0.9869	0.7723	0.7710	0.7717
0.1392	15.0	345	0.9069	0.7831	0.7812	0.7822
0.1318	16.0	368	1.0261	0.7669	0.7666	0.7668
0.1167	17.0	391	1.0508	0.7623	0.7635	0.7629
0.1168	18.0	414	1.1379	0.7509	0.7501	0.7505
0.1097	19.0	437	1.0633	0.7691	0.7717	0.7704
0.104	20.0	460	1.0595	0.7729	0.7741	0.7735

Framework versions

• Transformers 4.46.3
• Pytorch 2.2.1+cu118
• Datasets 2.17.0
• Tokenizers 0.20.3