hubert-base-superb-er-3kfoldfull15-finetuned-bmd-20250824_100720-LOSO-section-out1

This model is a fine-tuned version of superb/hubert-base-superb-er on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0314
• Accuracy: 0.5
• F1: 0.4905

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 1968
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	1.0	3	1.1067	0.3333	0.1667
No log	2.0	6	1.1031	0.375	0.2461
No log	3.0	9	1.0992	0.1667	0.125
1.0983	4.0	12	1.0958	0.3333	0.2626
1.0983	5.0	15	1.0929	0.2917	0.2536
1.0983	6.0	18	1.0903	0.3333	0.2296
1.0799	7.0	21	1.0874	0.3333	0.2296
1.0799	8.0	24	1.0820	0.3333	0.2296
1.0799	9.0	27	1.0754	0.3333	0.2775
1.057	10.0	30	1.0664	0.3333	0.2775
1.057	11.0	33	1.0563	0.3333	0.2775
1.057	12.0	36	1.0450	0.375	0.3333
1.057	13.0	39	1.0377	0.4167	0.3983
1.0304	14.0	42	1.0334	0.5	0.4905
1.0304	15.0	45	1.0314	0.5	0.4905

Framework versions

• Transformers 4.55.2
• Pytorch 2.8.0+cu126
• Datasets 3.6.0
• Tokenizers 0.21.4