wavlm-malayalam-binary

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

• Loss: 0.2800
• Accuracy: 0.9659
• F1: 0.9677

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: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.6631	1.0	23	0.6537	0.8409	0.8478
0.4699	2.0	46	0.2677	0.9205	0.9247
0.2441	3.0	69	0.2378	0.9318	0.9333
0.1424	4.0	92	0.1722	0.9545	0.9565
0.097	5.0	115	0.1579	0.9659	0.9691
0.1097	6.0	138	0.1662	0.9659	0.9677
0.04	7.0	161	0.1843	0.9659	0.9677
0.0591	8.0	184	0.2041	0.9659	0.9677
0.0262	9.0	207	0.2261	0.9659	0.9677
0.0249	10.0	230	0.3226	0.9545	0.9565
0.0401	11.0	253	0.1761	0.9773	0.9787
0.0199	12.0	276	0.1896	0.9773	0.9787
0.012	13.0	299	0.2001	0.9773	0.9787
0.0128	14.0	322	0.2171	0.9773	0.9787
0.0289	15.0	345	0.2275	0.9773	0.9787
0.0001	16.0	368	0.2367	0.9773	0.9787
0.0197	17.0	391	0.2454	0.9773	0.9787
0.0083	18.0	414	0.3432	0.9659	0.9677
0.0002	19.0	437	0.3327	0.9659	0.9677
0.0018	20.0	460	0.2800	0.9659	0.9677

Framework versions

• Transformers 4.53.3
• Pytorch 2.9.1+cu128
• Datasets 4.4.1
• Tokenizers 0.21.2