hbertv1-Massive-intent

This model is a fine-tuned version of gokuls/bert_12_layer_model_v1_complete_training_new on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8959
• Accuracy: 0.8451

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
3.051	1.0	180	1.8409	0.4968
1.3906	2.0	360	1.0234	0.7167
0.8613	3.0	540	0.8787	0.7688
0.6447	4.0	720	0.8405	0.7811
0.4955	5.0	900	0.8426	0.7850
0.3899	6.0	1080	0.7777	0.8175
0.3052	7.0	1260	0.7779	0.8175
0.2413	8.0	1440	0.8294	0.8254
0.196	9.0	1620	0.8265	0.8214
0.1545	10.0	1800	0.8361	0.8362
0.1177	11.0	1980	0.8470	0.8288
0.0894	12.0	2160	0.8706	0.8283
0.0666	13.0	2340	0.8853	0.8392
0.0447	14.0	2520	0.8959	0.8451
0.0312	15.0	2700	0.8982	0.8441

Framework versions

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