hBERTv2_new_pretrain_w_init__qnli

This model is a fine-tuned version of gokuls/bert_12_layer_model_v2_complete_training_new_wt_init on the GLUE QNLI dataset. It achieves the following results on the evaluation set:

• Loss: 0.6573
• Accuracy: 0.5958

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
0.6809	1.0	819	0.6640	0.6021
0.6539	2.0	1638	0.6573	0.5958
0.6358	3.0	2457	0.6694	0.5942
0.6313	4.0	3276	0.6898	0.5790
0.6293	5.0	4095	0.6801	0.5880
0.6253	6.0	4914	0.6829	0.5925
0.6166	7.0	5733	0.6865	0.5916

Framework versions

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