hbertv1-emotion-logit_KD-tiny

This model is a fine-tuned version of gokuls/model_v1_complete_training_wt_init_48_tiny_freeze_new on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.4386
• Accuracy: 0.8995

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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.9341	1.0	250	2.0281	0.6225
1.5579	2.0	500	1.0162	0.812
0.9088	3.0	750	0.6563	0.8705
0.6557	4.0	1000	0.5484	0.879
0.538	5.0	1250	0.4913	0.8865
0.4524	6.0	1500	0.4836	0.888
0.4072	7.0	1750	0.4416	0.896
0.3797	8.0	2000	0.4346	0.8905
0.3426	9.0	2250	0.4386	0.8995
0.3183	10.0	2500	0.4602	0.896
0.2911	11.0	2750	0.4296	0.8945
0.2807	12.0	3000	0.4442	0.896
0.2609	13.0	3250	0.4513	0.894
0.249	14.0	3500	0.4612	0.8975

Framework versions

• Transformers 4.35.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.15.0
• Tokenizers 0.15.0