model_broadclass_onSet1

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9014
• 0 Precision: 0.5217
• 0 Recall: 1.0
• 0 F1-score: 0.6857
• 0 Support: 24
• 1 Precision: 1.0
• 1 Recall: 0.7692
• 1 F1-score: 0.8696
• 1 Support: 39
• 2 Precision: 1.0
• 2 Recall: 0.5652
• 2 F1-score: 0.7222
• 2 Support: 23
• 3 Precision: 1.0
• 3 Recall: 0.75
• 3 F1-score: 0.8571
• 3 Support: 12
• Accuracy: 0.7755
• Macro avg Precision: 0.8804
• Macro avg Recall: 0.7711
• Macro avg F1-score: 0.7837
• Macro avg Support: 98
• Weighted avg Precision: 0.8829
• Weighted avg Recall: 0.7755
• Weighted avg F1-score: 0.7884
• Weighted avg Support: 98
• Wer: 0.9368
• Mtrix: [[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 9, 30, 0, 0], [2, 10, 0, 13, 0], [3, 3, 0, 0, 9]]

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 70
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	0 Precision	0 Recall	0 F1-score	0 Support	1 Precision	1 Recall	1 F1-score	1 Support	2 Precision	2 Recall	2 F1-score	2 Support	3 Precision	3 Recall	3 F1-score	3 Support	Accuracy	Macro avg Precision	Macro avg Recall	Macro avg F1-score	Macro avg Support	Weighted avg Precision	Weighted avg Recall	Weighted avg F1-score	Weighted avg Support	Wer	Mtrix
2.395	4.16	100	2.2004	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
2.2919	8.33	200	2.1576	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
2.0987	12.49	300	2.0882	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.9079	16.65	400	1.8619	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.7168	20.82	500	1.6469	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.551	24.98	600	1.6614	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.6399	29.16	700	1.5818	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.3329	33.33	800	1.2267	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.1996	37.49	900	1.2143	0.2449	1.0	0.3934	24	0.0	0.0	0.0	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2449	0.0612	0.25	0.0984	98	0.0600	0.2449	0.0964	98	0.9879	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 39, 0, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
1.01	41.65	1000	0.9496	0.2474	1.0	0.3967	24	1.0	0.0256	0.05	39	0.0	0.0	0.0	23	0.0	0.0	0.0	12	0.2551	0.3119	0.2564	0.1117	98	0.4586	0.2551	0.1170	98	0.9777	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 38, 1, 0, 0], [2, 23, 0, 0, 0], [3, 12, 0, 0, 0]]
0.9516	45.82	1100	0.9471	0.2927	1.0	0.4528	24	1.0	0.3846	0.5556	39	1.0	0.0435	0.0833	23	0.0	0.0	0.0	12	0.4082	0.5732	0.3570	0.2729	98	0.7043	0.4082	0.3515	98	0.9661	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 24, 15, 0, 0], [2, 22, 0, 1, 0], [3, 12, 0, 0, 0]]
0.9544	49.98	1200	0.9452	0.3582	1.0	0.5275	24	1.0	0.5128	0.6780	39	1.0	0.3043	0.4667	23	0.75	0.25	0.375	12	0.5510	0.7771	0.5168	0.5118	98	0.8122	0.5510	0.5544	98	0.9540	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 18, 20, 0, 1], [2, 16, 0, 7, 0], [3, 9, 0, 0, 3]]
0.9538	54.16	1300	0.9259	0.4615	1.0	0.6316	24	1.0	0.6923	0.8182	39	1.0	0.5217	0.6857	23	0.8571	0.5	0.6316	12	0.7041	0.8297	0.6785	0.6918	98	0.8506	0.7041	0.7185	98	0.9439	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 11, 27, 0, 1], [2, 11, 0, 12, 0], [3, 6, 0, 0, 6]]
0.952	58.33	1400	0.9052	0.4528	1.0	0.6234	24	1.0	0.6667	0.8	39	1.0	0.4348	0.6061	23	0.8889	0.6667	0.7619	12	0.6939	0.8354	0.6920	0.6978	98	0.8524	0.6939	0.7066	98	0.9464	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 12, 26, 0, 1], [2, 13, 0, 10, 0], [3, 4, 0, 0, 8]]
0.8938	62.49	1500	0.9070	0.48	1.0	0.6486	24	0.9677	0.7692	0.8571	39	1.0	0.4348	0.6061	23	1.0	0.5833	0.7368	12	0.7245	0.8619	0.6968	0.7122	98	0.8598	0.7245	0.7324	98	0.9398	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 9, 30, 0, 0], [2, 12, 1, 10, 0], [3, 5, 0, 0, 7]]
0.9027	66.65	1600	0.8919	0.5714	1.0	0.7273	24	1.0	0.8462	0.9167	39	1.0	0.7391	0.85	23	1.0	0.5	0.6667	12	0.8163	0.8929	0.7713	0.7902	98	0.8950	0.8163	0.8240	98	0.9398	[[0, 1, 2, 3], [0, 24, 0, 0, 0], [1, 6, 33, 0, 0], [2, 6, 0, 17, 0], [3, 6, 0, 0, 6]]

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.0+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2