Wav2Vec2_EmoRecog_Model_v5

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

• Loss: 1.9448
• Accuracy: 0.4436

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: 1.5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.7312	1.0	753	1.7018	0.3267
1.6018	2.0	1506	1.5855	0.3738
1.5244	3.0	2259	1.5603	0.3931
1.4636	4.0	3012	1.5217	0.4017
1.3947	5.0	3765	1.5199	0.4309
1.3283	6.0	4518	1.5389	0.4230
1.2596	7.0	5271	1.5628	0.4250
1.191	8.0	6024	1.5612	0.4555
1.1193	9.0	6777	1.5749	0.4542
1.0812	10.0	7530	1.6029	0.4462
1.0058	11.0	8283	1.6582	0.4422
0.9441	12.0	9036	1.7178	0.4290
0.8932	13.0	9789	1.7609	0.4502
0.8477	14.0	10542	1.8105	0.4469
0.7997	15.0	11295	1.7957	0.4382
0.7716	16.0	12048	1.8160	0.4502
0.7133	17.0	12801	1.9470	0.4329
0.7034	18.0	13554	1.9191	0.4462
0.6867	19.0	14307	1.9275	0.4495
0.6435	20.0	15060	1.9448	0.4436

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0