hbertv2-Massive-intent_48_KD

This model is a fine-tuned version of gokuls/bert_12_layer_model_v2_complete_training_new_48_KD on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8350
• Accuracy: 0.8578

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
2.0643	1.0	180	1.1020	0.7118
0.9395	2.0	360	0.8743	0.7600
0.6824	3.0	540	0.7820	0.8087
0.528	4.0	720	0.7761	0.8057
0.4245	5.0	900	0.7209	0.8333
0.3374	6.0	1080	0.7113	0.8337
0.2666	7.0	1260	0.7220	0.8362
0.2107	8.0	1440	0.7670	0.8377
0.1735	9.0	1620	0.7526	0.8411
0.1276	10.0	1800	0.8256	0.8465
0.0984	11.0	1980	0.8074	0.8495
0.0697	12.0	2160	0.7939	0.8564
0.0472	13.0	2340	0.8350	0.8578
0.0292	14.0	2520	0.8511	0.8554
0.018	15.0	2700	0.8536	0.8574

Framework versions

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