model_phoneme_onSet2

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: 0.1464
• 0 Precision: 1.0
• 0 Recall: 0.9677
• 0 F1-score: 0.9836
• 0 Support: 31
• 1 Precision: 0.9167
• 1 Recall: 1.0
• 1 F1-score: 0.9565
• 1 Support: 22
• 2 Precision: 1.0
• 2 Recall: 0.9333
• 2 F1-score: 0.9655
• 2 Support: 30
• 3 Precision: 0.9333
• 3 Recall: 1.0
• 3 F1-score: 0.9655
• 3 Support: 14
• Accuracy: 0.9691
• Macro avg Precision: 0.9625
• Macro avg Recall: 0.9753
• Macro avg F1-score: 0.9678
• Macro avg Support: 97
• Weighted avg Precision: 0.9715
• Weighted avg Recall: 0.9691
• Weighted avg F1-score: 0.9693
• Weighted avg Support: 97
• Wer: 0.1380
• Mtrix: [[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 28, 1], [3, 0, 0, 0, 14]]

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
3.9429	4.16	100	3.3748	0.0	0.0	0.0	31	0.0	0.0	0.0	22	0.0	0.0	0.0	30	0.1011	0.6429	0.1748	14	0.0928	0.0253	0.1607	0.0437	97	0.0146	0.0928	0.0252	97	0.9980	[[0, 1, 2, 3], [0, 0, 0, 0, 31], [1, 0, 0, 0, 22], [2, 3, 0, 0, 27], [3, 5, 0, 0, 9]]
3.3504	8.33	200	3.1724	0.0	0.0	0.0	31	0.0	0.0	0.0	22	0.0	0.0	0.0	30	0.1011	0.6429	0.1748	14	0.0928	0.0253	0.1607	0.0437	97	0.0146	0.0928	0.0252	97	0.9980	[[0, 1, 2, 3], [0, 0, 0, 0, 31], [1, 0, 0, 0, 22], [2, 3, 0, 0, 27], [3, 5, 0, 0, 9]]
3.155	12.49	300	3.1448	0.0	0.0	0.0	31	0.0	0.0	0.0	22	0.0	0.0	0.0	30	0.1011	0.6429	0.1748	14	0.0928	0.0253	0.1607	0.0437	97	0.0146	0.0928	0.0252	97	0.9980	[[0, 1, 2, 3], [0, 0, 0, 0, 31], [1, 0, 0, 0, 22], [2, 3, 0, 0, 27], [3, 5, 0, 0, 9]]
3.0282	16.65	400	2.9990	0.0	0.0	0.0	31	0.2268	1.0	0.3697	22	0.0	0.0	0.0	30	0.0	0.0	0.0	14	0.2268	0.0567	0.25	0.0924	97	0.0514	0.2268	0.0839	97	1.0	[[0, 1, 2, 3], [0, 0, 31, 0, 0], [1, 0, 22, 0, 0], [2, 0, 30, 0, 0], [3, 0, 14, 0, 0]]
2.744	20.82	500	2.6658	0.8378	1.0	0.9118	31	0.3889	0.6364	0.4828	22	0.4583	0.3667	0.4074	30	0.0	0.0	0.0	14	0.5773	0.4213	0.5008	0.4505	97	0.4977	0.5773	0.5269	97	1.0	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 6, 14, 2, 0], [2, 0, 19, 11, 0], [3, 0, 3, 11, 0]]
2.2503	24.98	600	2.0915	0.9677	0.9677	0.9677	31	0.8571	0.8182	0.8372	22	0.875	0.9333	0.9032	30	0.9231	0.8571	0.8889	14	0.9072	0.9057	0.8941	0.8993	97	0.9075	0.9072	0.9068	97	0.9609	[[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 1, 18, 2, 1], [2, 0, 2, 28, 0], [3, 0, 0, 2, 12]]
1.8687	29.16	700	1.7109	1.0	0.9355	0.9667	31	0.7857	1.0	0.88	22	1.0	0.9333	0.9655	30	1.0	0.8571	0.9231	14	0.9381	0.9464	0.9315	0.9338	97	0.9514	0.9381	0.9404	97	0.9373	[[0, 1, 2, 3], [0, 29, 2, 0, 0], [1, 0, 22, 0, 0], [2, 0, 2, 28, 0], [3, 0, 2, 0, 12]]
1.4444	33.33	800	1.3295	1.0	0.9677	0.9836	31	0.88	1.0	0.9362	22	1.0	0.9667	0.9831	30	1.0	0.9286	0.9630	14	0.9691	0.97	0.9657	0.9664	97	0.9728	0.9691	0.9697	97	0.9142	[[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 29, 0], [3, 0, 1, 0, 13]]
0.95	37.49	900	0.8782	1.0	1.0	1.0	31	0.9167	1.0	0.9565	22	1.0	0.9333	0.9655	30	0.9286	0.9286	0.9286	14	0.9691	0.9613	0.9655	0.9627	97	0.9708	0.9691	0.9692	97	0.8545	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 28, 1], [3, 0, 1, 0, 13]]
0.5303	41.65	1000	0.4750	1.0	1.0	1.0	31	0.9167	1.0	0.9565	22	1.0	0.9333	0.9655	30	1.0	1.0	1.0	14	0.9794	0.9792	0.9833	0.9805	97	0.9811	0.9794	0.9795	97	0.6026	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 0, 22, 0, 0], [2, 0, 2, 28, 0], [3, 0, 0, 0, 14]]
0.3054	45.82	1100	0.2919	0.9688	1.0	0.9841	31	0.9130	0.9545	0.9333	22	1.0	0.9	0.9474	30	0.9333	1.0	0.9655	14	0.9588	0.9538	0.9636	0.9576	97	0.9607	0.9588	0.9586	97	0.2373	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 1, 21, 0, 0], [2, 0, 2, 27, 1], [3, 0, 0, 0, 14]]
0.1609	49.98	1200	0.1727	1.0	1.0	1.0	31	0.88	1.0	0.9362	22	1.0	0.9	0.9474	30	0.9286	0.9286	0.9286	14	0.9588	0.9521	0.9571	0.9530	97	0.9625	0.9588	0.9589	97	0.1646	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 0, 22, 0, 0], [2, 0, 2, 27, 1], [3, 0, 1, 0, 13]]
0.1204	54.16	1300	0.1430	1.0	1.0	1.0	31	0.9167	1.0	0.9565	22	1.0	0.9333	0.9655	30	0.9286	0.9286	0.9286	14	0.9691	0.9613	0.9655	0.9627	97	0.9708	0.9691	0.9692	97	0.1370	[[0, 1, 2, 3], [0, 31, 0, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 28, 1], [3, 0, 1, 0, 13]]
0.0924	58.33	1400	0.1494	0.9677	0.9677	0.9677	31	0.9130	0.9545	0.9333	22	1.0	0.9333	0.9655	30	0.9333	1.0	0.9655	14	0.9588	0.9535	0.9639	0.9580	97	0.9603	0.9588	0.9589	97	0.1581	[[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 1, 21, 0, 0], [2, 0, 1, 28, 1], [3, 0, 0, 0, 14]]
0.0596	62.49	1500	0.1484	1.0	0.9677	0.9836	31	0.9167	1.0	0.9565	22	1.0	0.9333	0.9655	30	0.9333	1.0	0.9655	14	0.9691	0.9625	0.9753	0.9678	97	0.9715	0.9691	0.9693	97	0.1370	[[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 28, 1], [3, 0, 0, 0, 14]]
0.0592	66.65	1600	0.1464	1.0	0.9677	0.9836	31	0.9167	1.0	0.9565	22	1.0	0.9333	0.9655	30	0.9333	1.0	0.9655	14	0.9691	0.9625	0.9753	0.9678	97	0.9715	0.9691	0.9693	97	0.1380	[[0, 1, 2, 3], [0, 30, 1, 0, 0], [1, 0, 22, 0, 0], [2, 0, 1, 28, 1], [3, 0, 0, 0, 14]]

Framework versions

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