hBERTv2_new_pretrain_48_KD_cola

This model is a fine-tuned version of gokuls/bert_12_layer_model_v2_complete_training_new_48_KD on the GLUE COLA dataset. It achieves the following results on the evaluation set:

• Loss: 0.6132
• Matthews Correlation: 0.0647
• Accuracy: 0.6779

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: 4e-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	Matthews Correlation	Accuracy
0.6252	1.0	67	0.6183	0.0	0.6913
0.6155	2.0	134	0.6184	0.0	0.6913
0.6098	3.0	201	0.6132	0.0647	0.6779
0.6009	4.0	268	0.6308	0.0985	0.6779
0.5755	5.0	335	0.6287	0.0766	0.6894
0.5478	6.0	402	0.6476	0.0967	0.6625
0.5158	7.0	469	0.7149	0.0667	0.6539
0.4864	8.0	536	0.6457	0.0789	0.6500

Framework versions

• Transformers 4.30.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.12.0
• Tokenizers 0.13.3