iewav2vec2-finetuned-on-shEMO

This model is a fine-tuned version of minoosh/wav2vec2-base-finetuned-ie on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9088
• Accuracy: 0.82

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.0779	1.0	75	0.9736	0.7133
0.7753	2.0	150	0.7869	0.7467
0.5784	3.0	225	0.6579	0.7867
0.5088	4.0	300	0.6393	0.7933
0.4454	5.0	375	0.6090	0.83
0.3197	6.0	450	0.7735	0.8067
0.235	7.0	525	0.6943	0.8333
0.1648	8.0	600	0.6958	0.8367
0.1016	9.0	675	0.7437	0.8333
0.0822	10.0	750	0.6435	0.8467
0.1434	11.0	825	0.6448	0.86
0.0748	12.0	900	0.7208	0.86
0.0403	13.0	975	0.6549	0.8733
0.0365	14.0	1050	0.7136	0.8667
0.0249	15.0	1125	0.6958	0.87
0.0462	16.0	1200	0.8340	0.8533
0.0061	17.0	1275	0.8503	0.84
0.0257	18.0	1350	0.9144	0.85
0.0262	19.0	1425	0.8671	0.8567
0.005	20.0	1500	0.8572	0.8567

Framework versions

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