HBERTv1_48_L6_H768_A12_massive

This model is a fine-tuned version of gokuls/HBERTv1_48_L6_H768_A12 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8228
• Accuracy: 0.8628

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
1.8011	1.0	180	0.8508	0.7678
0.712	2.0	360	0.7073	0.8121
0.463	3.0	540	0.6127	0.8426
0.3124	4.0	720	0.6141	0.8416
0.2196	5.0	900	0.6738	0.8431
0.1592	6.0	1080	0.7337	0.8387
0.1209	7.0	1260	0.7453	0.8411
0.0803	8.0	1440	0.7655	0.8460
0.0552	9.0	1620	0.7871	0.8495
0.0452	10.0	1800	0.7842	0.8598
0.027	11.0	1980	0.8228	0.8628
0.0149	12.0	2160	0.8452	0.8564
0.0085	13.0	2340	0.8708	0.8554
0.0052	14.0	2520	0.8564	0.8623
0.0033	15.0	2700	0.8652	0.8588

Framework versions

• Transformers 4.34.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.14.5
• Tokenizers 0.14.0