hBERTv2_data_aug_mnli

This model is a fine-tuned version of gokuls/bert_12_layer_model_v2 on the GLUE MNLI dataset. It achieves the following results on the evaluation set:

• Loss: 1.0988
• Accuracy: 0.3182

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
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.0988	1.0	31440	1.0988	0.3182
1.0985	2.0	62880	1.0992	0.3182
1.0985	3.0	94320	1.0991	0.3182
1.0985	4.0	125760	1.0991	0.3182
1.0985	5.0	157200	1.0988	0.3182
1.0985	6.0	188640	1.0988	0.3182
1.0985	7.0	220080	1.0988	0.3182
1.0985	8.0	251520	1.0988	0.3182
1.0985	9.0	282960	1.0988	0.3182
1.0985	10.0	314400	1.0988	0.3182
1.0985	11.0	345840	1.0988	0.3182
1.0985	12.0	377280	1.0988	0.3182
1.0985	13.0	408720	1.0988	0.3182
1.0985	14.0	440160	1.0988	0.3182
1.0985	15.0	471600	1.0988	0.3182

Framework versions

• Transformers 4.26.1
• Pytorch 1.14.0a0+410ce96
• Datasets 2.10.1
• Tokenizers 0.13.2