HBERTv1_48_L10_H128_A2_emotion

This model is a fine-tuned version of gokuls/HBERTv1_48_L10_H128_A2 on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.3362
• Accuracy: 0.8865

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
1.4132	1.0	250	1.1283	0.5875
0.9519	2.0	500	0.7405	0.757
0.6375	3.0	750	0.5533	0.8295
0.4709	4.0	1000	0.4480	0.8625
0.3802	5.0	1250	0.4056	0.8665
0.3246	6.0	1500	0.3581	0.877
0.2718	7.0	1750	0.3616	0.877
0.2422	8.0	2000	0.3427	0.8805
0.2157	9.0	2250	0.3452	0.8845
0.2026	10.0	2500	0.3362	0.8865

Framework versions

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