finetuned_wav2vec2.0-base-on-IEMOCAP_3

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: 1.3119
• Accuracy: 0.6412

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
1.197	1.0	109	1.1517	0.4738
0.9312	2.0	219	0.9722	0.5774
0.8183	3.0	328	0.8322	0.6640
0.7456	4.0	438	0.7201	0.7198
0.599	5.0	547	0.7853	0.7118
0.4166	6.0	657	0.6359	0.7711
0.2507	7.0	766	0.7845	0.7415
0.2794	8.0	876	0.8188	0.7494
0.1777	9.0	985	0.7406	0.7950
0.1344	10.0	1095	0.8735	0.7620
0.0719	11.0	1204	0.9456	0.7688
0.0577	12.0	1314	0.9727	0.7768
0.0405	13.0	1423	1.0440	0.7756
0.0619	14.0	1533	1.0239	0.7813
0.0299	14.93	1635	1.0343	0.7790

Framework versions

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