emotion_detection_model

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: 0.6542
• Accuracy: 0.8291

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.6124	0.99	70	1.5873	0.3984
1.2504	1.99	141	1.2053	0.5963
0.833	3.0	212	0.8178	0.7504
0.6633	4.0	283	0.7137	0.7783
0.5791	4.99	353	0.6395	0.7915
0.4472	5.99	424	0.6398	0.7968
0.378	7.0	495	0.5669	0.8145
0.2902	8.0	566	0.5777	0.8158
0.2621	8.99	636	0.6320	0.8074
0.231	9.99	707	0.6347	0.8149
0.174	11.0	778	0.6649	0.8096
0.1781	12.0	849	0.6180	0.8211
0.1566	12.99	919	0.6311	0.8211
0.1239	13.99	990	0.6322	0.8207
0.1223	15.0	1061	0.6443	0.8264
0.0988	16.0	1132	0.6424	0.8255
0.0866	16.99	1202	0.6542	0.8291
0.0661	17.99	1273	0.6748	0.8264
0.0815	19.0	1344	0.6723	0.8286
0.0595	19.79	1400	0.6865	0.8229

Framework versions

• Transformers 4.37.0
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.1