hubert-emotion-classifier-pt-en-v1

This model is a fine-tuned version of facebook/hubert-base-ls960 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7563
• Accuracy: 0.7781
• F1: 0.7804

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
2.043	1.0	507	1.8849	0.2732	0.1529
1.8343	2.0	1014	1.6340	0.3609	0.2655
1.6091	3.0	1521	1.4294	0.4862	0.4604
1.382	4.0	2028	1.1935	0.6045	0.5882
1.1837	5.0	2535	1.0617	0.6607	0.6545
1.026	6.0	3042	0.9764	0.6775	0.6753
0.9294	7.0	3549	0.8574	0.7308	0.7311
0.8115	8.0	4056	0.8209	0.7268	0.7257
0.7309	9.0	4563	0.7914	0.7426	0.7437
0.6686	10.0	5070	0.8395	0.7436	0.7404
0.6102	11.0	5577	0.7574	0.7673	0.7703
0.5866	12.0	6084	0.8018	0.7416	0.7440
0.5242	13.0	6591	0.7226	0.7791	0.7817
0.4732	14.0	7098	0.7636	0.7682	0.7723
0.4755	15.0	7605	0.7563	0.7781	0.7804

Framework versions

• Transformers 4.53.1
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.2