bert_12_layer_model_v3_48_massive

This model is a fine-tuned version of gokuls/bert_12_layer_model_v3_complete_training_48 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.9214
• Accuracy: 0.8608

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: 64
• eval_batch_size: 64
• seed: 33
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.7608	1.0	180	0.9351	0.7482
0.8417	2.0	360	0.8314	0.7841
0.6057	3.0	540	0.7307	0.8160
0.4715	4.0	720	0.6995	0.8382
0.3766	5.0	900	0.7602	0.8273
0.2933	6.0	1080	0.7537	0.8357
0.2321	7.0	1260	0.7966	0.8426
0.1873	8.0	1440	0.8015	0.8396
0.1447	9.0	1620	0.8281	0.8392
0.1079	10.0	1800	0.8665	0.8446
0.0781	11.0	1980	0.8758	0.8500
0.0573	12.0	2160	0.8646	0.8515
0.0383	13.0	2340	0.9121	0.8603
0.024	14.0	2520	0.8963	0.8593
0.015	15.0	2700	0.9214	0.8608

Framework versions

• Transformers 4.34.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.14.5
• Tokenizers 0.14.1