hBERTv1_new_pretrain_rte

This model is a fine-tuned version of gokuls/bert_12_layer_model_v1_complete_training_new on the GLUE RTE dataset. It achieves the following results on the evaluation set:

• Loss: 0.6896
• Accuracy: 0.5307

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	Accuracy
0.7407	1.0	20	0.7002	0.4729
0.7061	2.0	40	0.7245	0.4729
0.7102	3.0	60	0.6949	0.5271
0.703	4.0	80	0.6951	0.4729
0.7097	5.0	100	0.6974	0.4729
0.7006	6.0	120	0.7053	0.4729
0.6986	7.0	140	0.6896	0.5307
0.6935	8.0	160	0.7711	0.4729
0.6109	9.0	180	0.8443	0.4982
0.469	10.0	200	1.0369	0.5126
0.3028	11.0	220	1.1621	0.5235
0.2155	12.0	240	1.2096	0.5379

Framework versions

• Transformers 4.29.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.12.0
• Tokenizers 0.13.3