hbertv1-Massive-intent_48_emb_compress

This model is a fine-tuned version of gokuls/bert_12_layer_model_v1_complete_training_new_emb_compress_48 on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.9925
• Accuracy: 0.8392

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
2.2761	1.0	180	1.2513	0.6650
1.0561	2.0	360	0.9372	0.7442
0.7553	3.0	540	0.8482	0.7713
0.5586	4.0	720	0.8702	0.7737
0.4124	5.0	900	0.8478	0.7964
0.2983	6.0	1080	0.8568	0.8062
0.222	7.0	1260	0.8481	0.8175
0.1613	8.0	1440	0.8927	0.8091
0.1129	9.0	1620	0.9180	0.8195
0.085	10.0	1800	0.9829	0.8155
0.0517	11.0	1980	0.9875	0.8259
0.0302	12.0	2160	0.9917	0.8298
0.0169	13.0	2340	0.9807	0.8342
0.0073	14.0	2520	1.0070	0.8342
0.0032	15.0	2700	0.9925	0.8392

Framework versions

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