finetuned_wav2vec2.0-base-on-IEMOCAP_4

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6468
• Accuracy: 0.6586

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1816	1.0	112	1.1338	0.5177
1.1233	2.0	225	1.0683	0.5355
0.9672	3.0	337	0.9371	0.6031
0.6925	4.0	450	0.8086	0.6818
0.6162	5.0	562	0.7848	0.7051
0.4226	6.0	675	0.7377	0.7428
0.3047	7.0	787	0.7981	0.7251
0.2385	8.0	900	0.8558	0.7428
0.1954	9.0	1012	0.8869	0.7373
0.1454	10.0	1125	0.9470	0.7428
0.116	11.0	1237	1.0448	0.7306
0.1017	12.0	1350	1.1298	0.7295
0.1365	13.0	1462	1.1667	0.7450
0.0995	14.0	1575	1.1252	0.7428
0.0655	14.93	1680	1.1501	0.7395

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.0
• Datasets 2.12.0
• Tokenizers 0.13.3