bert_uncased_L-2_H-256_A-4_massive

This model is a fine-tuned version of google/bert_uncased_L-2_H-256_A-4 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8268
• Accuracy: 0.8062

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
3.6523	1.0	180	3.0957	0.3114
2.7875	2.0	360	2.3220	0.5352
2.1742	3.0	540	1.8439	0.6483
1.7765	4.0	720	1.5345	0.6940
1.4988	5.0	900	1.3275	0.7137
1.3009	6.0	1080	1.1805	0.7368
1.1512	7.0	1260	1.0746	0.7511
1.0374	8.0	1440	0.9977	0.7649
0.9466	9.0	1620	0.9426	0.7757
0.8821	10.0	1800	0.8991	0.7909
0.828	11.0	1980	0.8648	0.7929
0.7824	12.0	2160	0.8426	0.7988
0.7565	13.0	2340	0.8268	0.8062
0.7378	14.0	2520	0.8180	0.8052
0.7231	15.0	2700	0.8142	0.8047

Framework versions

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