HBERTv1_48_L6_H256_A4_massive

This model is a fine-tuned version of gokuls/HBERTv1_48_L6_H256_A4 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.7233
• Accuracy: 0.8460

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
3.3254	1.0	180	2.3694	0.4638
1.8706	2.0	360	1.3999	0.6616
1.2107	3.0	540	1.0206	0.7378
0.8953	4.0	720	0.8675	0.7821
0.6964	5.0	900	0.7948	0.7973
0.5749	6.0	1080	0.7426	0.8165
0.4668	7.0	1260	0.7449	0.8180
0.3947	8.0	1440	0.7142	0.8283
0.3345	9.0	1620	0.7030	0.8406
0.2859	10.0	1800	0.7111	0.8411
0.2418	11.0	1980	0.7323	0.8392
0.2145	12.0	2160	0.7269	0.8392
0.1885	13.0	2340	0.7233	0.8460
0.17	14.0	2520	0.7294	0.8411
0.1579	15.0	2700	0.7331	0.8436

Framework versions

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