bert_uncased_L-2_H-128_A-2_massive

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

• Loss: 1.6807
• Accuracy: 0.6734

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
3.8812	1.0	180	3.6329	0.2086
3.4819	2.0	360	3.2331	0.3074
3.1331	3.0	540	2.9025	0.3847
2.8406	4.0	720	2.6372	0.4624
2.6013	5.0	900	2.4211	0.5194
2.4097	6.0	1080	2.2485	0.5539
2.2504	7.0	1260	2.1084	0.5898
2.1234	8.0	1440	1.9968	0.6085
2.0195	9.0	1620	1.9036	0.6316
1.9345	10.0	1800	1.8336	0.6463
1.8616	11.0	1980	1.7722	0.6596
1.8091	12.0	2160	1.7281	0.6645
1.7704	13.0	2340	1.6984	0.6685
1.7431	14.0	2520	1.6807	0.6734
1.7293	15.0	2700	1.6749	0.6729

Framework versions

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