torgo_xlsr_finetune-F01-2

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: 2.0297
• Wer: 0.9501

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
23.5657	0.81	500	3.3629	1.0
3.349	1.62	1000	2.9797	1.0
2.6233	2.44	1500	1.9631	1.0317
1.6556	3.25	2000	1.5230	1.2971
1.1942	4.06	2500	1.3185	1.1814
0.9141	4.87	3000	1.3889	1.1723
0.7505	5.68	3500	1.2601	1.1247
0.6246	6.49	4000	1.4765	1.0907
0.5506	7.31	4500	1.3586	1.1020
0.5004	8.12	5000	1.2889	1.0454
0.4591	8.93	5500	1.6057	1.0680
0.4064	9.74	6000	1.4433	1.0023
0.3865	10.55	6500	1.7125	1.0567
0.3422	11.36	7000	1.7667	1.0045
0.3084	12.18	7500	1.6708	1.0567
0.2904	12.99	8000	1.8301	1.0045
0.2826	13.8	8500	1.5566	0.9796
0.2707	14.61	9000	1.6200	0.9909
0.2499	15.42	9500	1.9019	1.0045
0.2292	16.23	10000	1.7184	0.9841
0.2245	17.05	10500	1.7977	0.9932
0.2156	17.86	11000	1.7847	0.9705
0.2152	18.67	11500	1.9312	0.9660
0.2003	19.48	12000	2.1059	0.9909
0.1908	20.29	12500	1.9738	0.9683
0.1638	21.1	13000	1.7796	0.9592
0.1606	21.92	13500	2.0548	0.9569
0.1569	22.73	14000	1.9577	0.9274
0.1458	23.54	14500	2.1357	0.9569
0.1599	24.35	15000	1.9791	0.9410
0.1489	25.16	15500	1.9031	0.9456
0.1473	25.97	16000	1.8522	0.9342
0.1167	26.79	16500	1.8953	0.9342
0.1177	27.6	17000	1.9675	0.9342
0.1268	28.41	17500	1.9905	0.9501
0.1136	29.22	18000	2.0297	0.9501

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 1.18.3
• Tokenizers 0.13.2