bert_uncased_L-10_H-512_A-8_massive

This model is a fine-tuned version of google/bert_uncased_L-10_H-512_A-8 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.6000
• Accuracy: 0.8869

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
2.4668	1.0	180	1.3269	0.7511
1.0706	2.0	360	0.7806	0.8411
0.6117	3.0	540	0.6024	0.8662
0.3953	4.0	720	0.5695	0.8692
0.2706	5.0	900	0.5476	0.8751
0.1872	6.0	1080	0.5291	0.8775
0.1345	7.0	1260	0.5455	0.8770
0.0982	8.0	1440	0.5429	0.8844
0.067	9.0	1620	0.5639	0.8854
0.0513	10.0	1800	0.5952	0.8775
0.0392	11.0	1980	0.5867	0.8834
0.0305	12.0	2160	0.6004	0.8785
0.0241	13.0	2340	0.5916	0.8849
0.0202	14.0	2520	0.6000	0.8869
0.0187	15.0	2700	0.6025	0.8859

Framework versions

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