bert_uncased_L-8_H-128_A-2_massive

This model is a fine-tuned version of google/bert_uncased_L-8_H-128_A-2 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 1.4687
• Accuracy: 0.7334

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
3.7982	1.0	180	3.4791	0.3335
3.287	2.0	360	3.0244	0.4299
2.8916	3.0	540	2.6675	0.5047
2.5836	4.0	720	2.3965	0.5839
2.3397	5.0	900	2.1824	0.6291
2.1423	6.0	1080	2.0132	0.6680
1.9823	7.0	1260	1.8773	0.6872
1.8543	8.0	1440	1.7682	0.6940
1.7528	9.0	1620	1.6818	0.7049
1.6612	10.0	1800	1.6191	0.7118
1.595	11.0	1980	1.5608	0.7226
1.5406	12.0	2160	1.5231	0.7270
1.4997	13.0	2340	1.4922	0.7309
1.4719	14.0	2520	1.4760	0.7329
1.4595	15.0	2700	1.4687	0.7334

Framework versions

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