bert_uncased_L-4_H-256_A-4_massive

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

• Loss: 0.7252
• Accuracy: 0.8362

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
3.5031	1.0	180	2.8542	0.4437
2.5403	2.0	360	2.0782	0.6394
1.928	3.0	540	1.6213	0.7118
1.542	4.0	720	1.3355	0.7526
1.2771	5.0	900	1.1556	0.7801
1.0852	6.0	1080	1.0223	0.7964
0.939	7.0	1260	0.9331	0.8047
0.8352	8.0	1440	0.8670	0.8146
0.7522	9.0	1620	0.8184	0.8190
0.6847	10.0	1800	0.7887	0.8254
0.6369	11.0	1980	0.7578	0.8254
0.5943	12.0	2160	0.7413	0.8323
0.5652	13.0	2340	0.7288	0.8328
0.5486	14.0	2520	0.7252	0.8362
0.5394	15.0	2700	0.7190	0.8357

Framework versions

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