yip-i/torgo_xlsr_finetune-M01-no-lm

torgo_xlsr_finetune-M01-2

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: 1.5763
• Wer: 0.9555

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
23.0716	0.9	500	3.3142	1.0
3.4092	1.8	1000	3.2440	1.0
2.9015	2.7	1500	2.8209	1.0
2.7211	3.6	2000	2.4913	1.2728
2.0884	4.5	2500	1.7817	1.4841
1.3426	5.41	3000	1.5117	1.4678
0.9866	6.31	3500	1.4760	1.3781
0.7874	7.21	4000	1.2179	1.2516
0.6424	8.11	4500	1.4501	1.2226
0.5505	9.01	5000	1.4132	1.3343
0.4709	9.91	5500	1.3289	1.1604
0.4358	10.81	6000	1.2615	1.1102
0.3892	11.71	6500	1.5597	1.1060
0.3602	12.61	7000	1.4205	1.1322
0.3298	13.51	7500	1.4411	1.1237
0.3184	14.41	8000	1.4017	1.1004
0.2954	15.32	8500	1.3428	1.0806
0.2745	16.22	9000	1.4793	1.0982
0.2533	17.12	9500	1.6004	1.1124
0.2378	18.02	10000	1.5802	1.0700
0.2234	18.92	10500	1.4462	1.0473
0.2147	19.82	11000	1.3814	1.0042
0.202	20.72	11500	1.5665	1.0226
0.1691	21.62	12000	1.4534	0.9958
0.1993	22.52	12500	1.4851	0.9894
0.1591	23.42	13000	1.3746	0.9746
0.1602	24.32	13500	1.4077	0.9710
0.1417	25.23	14000	1.5074	0.9668
0.1302	26.13	14500	1.5024	0.9456
0.1334	27.03	15000	1.4816	0.9541
0.1269	27.93	15500	1.5501	0.9541
0.1254	28.83	16000	1.5593	0.9527
0.12	29.73	16500	1.5763	0.9555

Framework versions

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