marconilabmak/Luganda_speech_to_intent_wav2vec_xlsr_ctc

Luganda_speech_to_intent_wav2vec_xlsr

This model is a fine-tuned version of KasuleTrevor/wav2vec2-large-xls-r-300m-lg-cv-130hr-v1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1398
• Accuracy: 0.9785

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.8838	1.0	131	1.4964	0.7701
1.2181	2.0	262	0.1181	0.9902
0.2092	3.0	393	0.0778	0.9892
0.1436	4.0	524	0.0734	0.9902
0.1248	5.0	655	0.0810	0.9892
0.1198	6.0	786	0.0707	0.9902
0.0888	7.0	917	0.0606	0.9913
0.0627	8.0	1048	0.0651	0.9902
0.0509	9.0	1179	0.0616	0.9913
0.0379	10.0	1310	0.0615	0.9913

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.0+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1