hBERTv2_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: 1.0983
• Accuracy: 0.3522

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
1.0992	1.0	1534	1.0996	0.3182
1.0988	2.0	3068	1.0988	0.3182
1.0987	3.0	4602	1.0987	0.3274
1.0986	4.0	6136	1.0987	0.3274
1.0987	5.0	7670	1.0984	0.3545
1.0987	6.0	9204	1.0986	0.3274
1.0986	7.0	10738	1.0986	0.3545
1.0987	8.0	12272	1.0986	0.3545
1.0986	9.0	13806	1.0984	0.3545
1.0986	10.0	15340	1.0983	0.3545
1.0987	11.0	16874	1.0986	0.3182
1.0987	12.0	18408	1.0984	0.3182
1.0986	13.0	19942	1.0983	0.3545
1.0986	14.0	21476	1.0984	0.3182
1.0986	15.0	23010	1.0986	0.3545

Framework versions

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