HBERTv1_48_L2_H64_A2_emotion

This model is a fine-tuned version of gokuls/HBERTv1_48_L2_H64_A2 on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.7737
• Accuracy: 0.7465

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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.6257	1.0	250	1.5119	0.476
1.4278	2.0	500	1.3417	0.511
1.2638	3.0	750	1.1761	0.547
1.1093	4.0	1000	1.0327	0.6155
0.9736	5.0	1250	0.9427	0.644
0.8866	6.0	1500	0.8741	0.6825
0.8203	7.0	1750	0.8358	0.713
0.7767	8.0	2000	0.7912	0.7395
0.7478	9.0	2250	0.7750	0.744
0.7268	10.0	2500	0.7737	0.7465

Framework versions

• Transformers 4.34.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.14.5
• Tokenizers 0.14.0