wav2vec2-large

This model is a fine-tuned version of facebook/wav2vec2-large on the galsenai/waxal_dataset dataset. It achieves the following results on the evaluation set:

Loss: 0.3413
Accuracy: 0.9443
Precision: 0.9780
F1: 0.9604

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: 3e-05
train_batch_size: 12
eval_batch_size: 12
seed: 0
gradient_accumulation_steps: 4
total_train_batch_size: 48
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 32.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	F1
4.6314	1.01	500	4.9165	0.0205	0.0028	0.0049
3.7739	2.02	1000	4.4491	0.0356	0.0750	0.0252
2.5035	3.04	1500	4.1429	0.1129	0.2672	0.1114
1.5633	4.05	2000	3.1973	0.3676	0.6598	0.3830
1.0538	5.06	2500	2.5479	0.5889	0.8417	0.6557
0.7422	6.07	3000	1.4494	0.7825	0.8921	0.8194
0.5762	7.08	3500	1.3168	0.7726	0.9277	0.8267
0.46	8.1	4000	0.8783	0.8564	0.9532	0.8982
0.4007	9.11	4500	0.7524	0.8738	0.9637	0.9137
0.3374	10.12	5000	0.6386	0.8852	0.9678	0.9221
0.3108	11.13	5500	0.5049	0.9106	0.9681	0.9373
0.2735	12.15	6000	0.6097	0.8905	0.9624	0.9226
0.2716	13.16	6500	0.4543	0.9000	0.9569	0.9206
0.2484	14.17	7000	0.3965	0.9272	0.9742	0.9489
0.228	15.18	7500	0.6807	0.8856	0.9777	0.9257
0.2307	16.19	8000	0.5219	0.9174	0.9802	0.9464
0.2169	17.21	8500	0.4630	0.9121	0.9677	0.9338
0.1997	18.22	9000	0.5152	0.9128	0.9740	0.9398
0.1921	19.23	9500	0.5105	0.9144	0.9867	0.9476
0.1825	20.24	10000	0.6302	0.9053	0.9832	0.9407
0.1786	21.25	10500	0.4602	0.9272	0.9813	0.9524
0.1671	22.27	11000	0.5443	0.9147	0.9794	0.9444
0.1623	23.28	11500	0.3413	0.9443	0.9780	0.9604
0.1595	24.29	12000	0.4478	0.9288	0.9813	0.9531
0.151	25.3	12500	0.4178	0.9360	0.9818	0.9571
0.1472	26.32	13000	0.4154	0.9356	0.9833	0.9578
0.1473	27.33	13500	0.4549	0.9318	0.9837	0.9561
0.131	28.34	14000	0.3574	0.9424	0.9845	0.9621
0.134	29.35	14500	0.4475	0.9333	0.9840	0.9568
0.1282	30.36	15000	0.4012	0.9382	0.9837	0.9591
0.1307	31.38	15500	0.3552	0.9428	0.9847	0.9624

Framework versions

Transformers 4.27.0.dev0
Pytorch 1.11.0+cu113
Datasets 2.9.1.dev0
Tokenizers 0.13.2

galsenai/wav2vec2-large-waxal-keyword-spotting

wav2vec2-large

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

为此模型做种