HBERTv1_48_L2_H768_A12_massive

This model is a fine-tuned version of gokuls/HBERTv1_48_L2_H768_A12 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.7845
• Accuracy: 0.8642

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
1.4964	1.0	180	0.6712	0.8087
0.5902	2.0	360	0.5767	0.8416
0.3724	3.0	540	0.5509	0.8510
0.2499	4.0	720	0.5592	0.8554
0.1719	5.0	900	0.5892	0.8529
0.118	6.0	1080	0.6567	0.8505
0.0849	7.0	1260	0.6597	0.8455
0.0656	8.0	1440	0.7050	0.8554
0.0456	9.0	1620	0.7098	0.8593
0.0314	10.0	1800	0.7583	0.8633
0.0213	11.0	1980	0.7845	0.8642
0.0174	12.0	2160	0.7764	0.8613
0.0112	13.0	2340	0.7723	0.8593
0.0076	14.0	2520	0.7828	0.8598
0.0062	15.0	2700	0.7825	0.8603

Framework versions

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