finetuned_wav2vec2.0-base-on-IEMOCAP_1

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: 1.6629
• Accuracy: 0.6673

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
1.2016	1.0	111	1.1443	0.5
1.0629	2.0	222	0.9839	0.5830
0.8111	2.99	333	0.9295	0.6233
0.7455	4.0	445	0.9119	0.6570
0.4654	5.0	556	0.7440	0.7186
0.3648	6.0	667	0.8359	0.7007
0.307	6.99	778	0.7964	0.7377
0.2823	8.0	890	0.8882	0.7321
0.1869	9.0	1001	1.0550	0.7186
0.2022	10.0	1112	1.0110	0.7231
0.0522	10.99	1223	1.1135	0.7265
0.058	12.0	1335	1.1275	0.7489
0.0465	13.0	1446	1.1479	0.7545
0.0321	14.0	1557	1.1885	0.7534
0.0331	14.97	1665	1.2301	0.7511

Framework versions

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