hbertv1-Massive-intent_48_w_in

This model is a fine-tuned version of gokuls/bert_12_layer_model_v1_complete_training_new_wt_init_48 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8264
• Accuracy: 0.8736

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
1.6907	1.0	180	0.8443	0.7777
0.7472	2.0	360	0.6977	0.8210
0.5222	3.0	540	0.6538	0.8352
0.3848	4.0	720	0.6461	0.8357
0.284	5.0	900	0.6195	0.8524
0.2051	6.0	1080	0.6218	0.8574
0.149	7.0	1260	0.6915	0.8495
0.1108	8.0	1440	0.7420	0.8574
0.0806	9.0	1620	0.7204	0.8549
0.0565	10.0	1800	0.7570	0.8603
0.0355	11.0	1980	0.7622	0.8677
0.0246	12.0	2160	0.8344	0.8647
0.0124	13.0	2340	0.8276	0.8682
0.0072	14.0	2520	0.8264	0.8736
0.0042	15.0	2700	0.8328	0.8736

Framework versions

• Transformers 4.30.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.13.0
• Tokenizers 0.13.3