Wav2Vec2_EmoRecog_Model_v2

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

• Loss: 1.8440
• Accuracy: 0.4349

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• 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
1.7698	1.0	377	1.6608	0.3513
1.6102	2.0	754	1.6074	0.3625
1.5556	3.0	1131	1.5894	0.3778
1.4899	4.0	1508	1.5643	0.3858
1.4322	5.0	1885	1.5250	0.4084
1.3737	6.0	2262	1.5445	0.4110
1.3217	7.0	2639	1.5287	0.4210
1.2686	8.0	3016	1.5635	0.4243
1.1999	9.0	3393	1.5674	0.4223
1.1511	10.0	3770	1.5881	0.4363
1.087	11.0	4147	1.6162	0.4177
1.0309	12.0	4524	1.6487	0.4296
0.9778	13.0	4901	1.7363	0.4210
0.9344	14.0	5278	1.7568	0.4210
0.9108	15.0	5655	1.7051	0.4416
0.8449	16.0	6032	1.7945	0.4329
0.8268	17.0	6409	1.7778	0.4402
0.7991	18.0	6786	1.7972	0.4382
0.7604	19.0	7163	1.8238	0.4276
0.7329	20.0	7540	1.8440	0.4349

Framework versions

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