wav2vec2-large-xlsr-53-english-finetuned-ravdess-v5

This model is a fine-tuned version of jonatasgrosman/wav2vec2-large-xlsr-53-english on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8443
• Accuracy: 0.7257

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: 5e-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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	9	2.0697	0.1424
2.0767	2.0	18	2.0585	0.2292
2.0642	3.0	27	2.0382	0.2118
2.0463	4.0	36	1.9870	0.2361
1.9984	5.0	45	1.8878	0.3160
1.8817	6.0	54	1.7381	0.3785
1.743	7.0	63	1.6483	0.4062
1.6047	8.0	72	1.5459	0.4340
1.4919	9.0	81	1.4229	0.4653
1.4067	10.0	90	1.3539	0.4479
1.4067	11.0	99	1.2584	0.5243
1.3039	12.0	108	1.2465	0.5243
1.2376	13.0	117	1.1980	0.5451
1.1504	14.0	126	1.1339	0.625
1.0479	15.0	135	1.1273	0.6007
0.9986	16.0	144	1.0976	0.6215
0.9289	17.0	153	1.0150	0.6528
0.9288	18.0	162	0.9629	0.6667
0.8092	19.0	171	0.9882	0.6528
0.7641	20.0	180	0.9357	0.6806
0.7641	21.0	189	0.9578	0.6840
0.7073	22.0	198	0.8655	0.6806
0.7277	23.0	207	1.0007	0.6632
0.6614	24.0	216	0.8399	0.7222
0.6571	25.0	225	0.8995	0.6875
0.6304	26.0	234	0.8523	0.7118
0.6298	27.0	243	0.8918	0.7049
0.5929	28.0	252	0.8510	0.7222
0.5915	29.0	261	0.8443	0.7257
0.5807	30.0	270	0.8536	0.7257

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3