hbertv1-emotion-logit_KD-mini

This model is a fine-tuned version of gokuls/model_v1_complete_training_wt_init_48_mini_freeze_new on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.4302
• Accuracy: 0.902

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
2.0557	1.0	250	0.9294	0.8275
0.7977	2.0	500	0.5722	0.876
0.531	3.0	750	0.5091	0.8805
0.4472	4.0	1000	0.4683	0.8935
0.39	5.0	1250	0.4489	0.8975
0.3432	6.0	1500	0.4714	0.895
0.3148	7.0	1750	0.4302	0.902
0.2859	8.0	2000	0.4388	0.8955
0.2635	9.0	2250	0.4317	0.9
0.2409	10.0	2500	0.4433	0.901
0.2249	11.0	2750	0.4413	0.89
0.2159	12.0	3000	0.4607	0.897

Framework versions

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