paraphrase-MiniLM-L12-v2-CoLA

This model is a fine-tuned version of sentence-transformers/paraphrase-MiniLM-L12-v2 on the GLUE COLA dataset. It achieves the following results on the evaluation set:

• Loss: 0.4636
• Matthews Correlation: 0.5057

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: 8e-05
• train_batch_size: 64
• eval_batch_size: 16
• seed: 30198
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 16.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Matthews Correlation
0.5747	1.0	67	0.5394	0.3455
0.5025	2.0	134	0.4999	0.4270
0.3698	3.0	201	0.4636	0.5057
0.2969	4.0	268	0.5309	0.4751
0.2275	5.0	335	0.6238	0.4775
0.1859	6.0	402	0.6315	0.4867
0.1517	7.0	469	0.7783	0.4695
0.1016	8.0	536	0.6762	0.4901
0.1017	9.0	603	0.7412	0.5046
0.0898	10.0	670	0.7719	0.4877
0.0527	11.0	737	0.8627	0.4955
0.0582	12.0	804	0.8986	0.4738
0.074	13.0	871	0.9469	0.4942
0.0508	14.0	938	0.9436	0.4918
0.024	15.0	1005	0.9391	0.4919
0.0458	16.0	1072	0.9375	0.4946

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.8.0
• Tokenizers 0.13.1