hbertv1-tiny-wt-48-Massive-intent

This model is a fine-tuned version of gokuls/model_v1_complete_training_wt_init_48_tiny on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.8676
• Accuracy: 0.7723

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
3.7161	1.0	180	3.1936	0.2499
2.8544	2.0	360	2.3660	0.4058
2.2122	3.0	540	1.8566	0.5430
1.7979	4.0	720	1.5269	0.6370
1.5083	5.0	900	1.3016	0.6911
1.3044	6.0	1080	1.1672	0.7098
1.1652	7.0	1260	1.0709	0.7270
1.0703	8.0	1440	1.0045	0.7432
0.996	9.0	1620	0.9595	0.7511
0.9323	10.0	1800	0.9276	0.7550
0.8832	11.0	1980	0.9183	0.7565
0.8521	12.0	2160	0.8953	0.7649
0.8246	13.0	2340	0.8829	0.7649
0.8072	14.0	2520	0.8676	0.7723
0.7947	15.0	2700	0.8657	0.7708

Framework versions

• Transformers 4.31.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.13.1
• Tokenizers 0.13.3