wavlm-basic_n-f-n_8batch_5sec_0.0001lr_unfrozen

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

• Loss: 1.0704
• Accuracy: 0.7333
• F1: 0.7308

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.003
• num_epochs: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
2.3031	0.98	24	2.3002	0.1667	0.1148
2.2766	2.0	49	2.2805	0.15	0.0930
2.2298	2.98	73	2.0679	0.2333	0.1421
1.9839	4.0	98	1.8757	0.25	0.1380
1.7495	4.98	122	1.5981	0.4	0.3370
1.5318	6.0	147	1.4640	0.45	0.3698
1.2765	6.98	171	1.3181	0.5167	0.4437
1.261	8.0	196	1.0905	0.5833	0.5429
1.078	8.98	220	1.0944	0.55	0.5244
0.9116	10.0	245	0.8228	0.6167	0.5603
0.8973	10.98	269	0.8632	0.5833	0.5266
0.8033	12.0	294	0.9061	0.65	0.6398
0.7183	12.98	318	0.8047	0.7	0.6877
0.7526	14.0	343	0.6695	0.7333	0.7176
0.6381	14.98	367	0.7510	0.7833	0.7788
0.5266	16.0	392	0.6154	0.8	0.7901
0.4485	16.98	416	0.8614	0.75	0.7359
0.5123	18.0	441	1.0848	0.65	0.6306
0.4094	18.98	465	0.6748	0.7667	0.7680
0.3114	20.0	490	0.7406	0.75	0.7389
0.2668	20.98	514	0.8419	0.75	0.7424

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3