multilingual_speech_to_intent_wav2vec_xlsr

This model is a fine-tuned version of KasuleTrevor/wav2vec2-large-xls-r-300m-lg-cv-130hr-v1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1493
• Accuracy: 0.9804
• Precision: 0.9813
• Recall: 0.9804
• F1: 0.9805

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: 0.0003
• 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: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.65	1.0	219	0.1235	0.9795	0.9799	0.9795	0.9795
0.2315	2.0	438	0.1033	0.9851	0.9854	0.9851	0.9852
0.2355	3.0	657	0.1331	0.9724	0.9740	0.9724	0.9724
0.1943	4.0	876	0.2951	0.9250	0.9304	0.9250	0.9245
0.1854	5.0	1095	0.5676	0.8931	0.9056	0.8931	0.8925
0.1499	6.0	1314	0.3552	0.9243	0.9344	0.9243	0.9240
0.1461	7.0	1533	0.2503	0.9441	0.9492	0.9441	0.9442
0.1407	8.0	1752	0.2951	0.9214	0.9269	0.9214	0.9212
0.116	9.0	1971	0.3022	0.9391	0.9425	0.9391	0.9390
0.1142	10.0	2190	0.2169	0.9483	0.9526	0.9483	0.9483
0.1064	11.0	2409	0.5370	0.9115	0.9171	0.9115	0.9111
0.1067	12.0	2628	1.1525	0.8259	0.8471	0.8259	0.8266

Framework versions

• Transformers 4.43.3
• Pytorch 2.1.0+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1