README.md

metadata

library_name: transformers
tags:
  - generated_from_trainer
datasets:
  - common_voice_17_0
metrics:
  - wer
model-index:
  - name: wav2vec2-gpt2-enc-dec
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_17_0
          type: common_voice_17_0
          config: cs
          split: train+validation
          args: cs
        metrics:
          - name: Wer
            type: wer
            value: 0.7337164750957854

wav2vec2-gpt2-enc-dec

This model is a fine-tuned version of on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3169
• Wer: 0.7337

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.08
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7683	0.2744	2000	0.7136	1.0624
0.6998	0.5488	4000	0.6563	1.1959
0.6798	0.8232	6000	0.6385	1.2165
0.6563	1.0975	8000	0.6162	1.1502
0.6385	1.3719	10000	0.6097	1.2121
0.6509	1.6463	12000	0.6009	1.1585
0.6365	1.9207	14000	0.5816	1.1171
0.6148	2.1951	16000	0.5811	1.1667
0.6081	2.4695	18000	0.5620	1.0605
0.6116	2.7439	20000	0.5596	1.1314
0.5875	3.0182	22000	0.5496	1.1081
0.5831	3.2926	24000	0.5347	1.0435
0.5798	3.5670	26000	0.5363	1.0736
0.6004	3.8414	28000	0.5234	1.0120
0.5525	4.1158	30000	0.5096	0.9784
0.5616	4.3902	32000	0.5074	0.9910
0.5623	4.6646	34000	0.5054	0.9849
0.549	4.9389	36000	0.5028	0.9962
0.5334	5.2133	38000	0.4876	0.9537
0.5482	5.4877	40000	0.4865	0.9414
0.5277	5.7621	42000	0.4886	0.9789
0.5126	6.0365	44000	0.4769	0.9494
0.5246	6.3109	46000	0.4690	0.9278
0.5255	6.5853	48000	0.4688	0.9455
0.5156	6.8597	50000	0.4597	0.9236
0.5135	7.1340	52000	0.4581	0.9245
0.5041	7.4084	54000	0.4488	0.9067
0.505	7.6828	56000	0.4456	0.9050
0.5151	7.9572	58000	0.4441	0.9053
0.4953	8.2316	60000	0.4374	0.8990
0.4815	8.5060	62000	0.4338	0.8987
0.506	8.7804	64000	0.4295	0.8897
0.4847	9.0547	66000	0.4273	0.8922
0.4763	9.3291	68000	0.4206	0.8782
0.4822	9.6035	70000	0.4180	0.8790
0.4728	9.8779	72000	0.4131	0.8673
0.4645	10.1523	74000	0.4086	0.8708
0.4696	10.4267	76000	0.4087	0.8741
0.4664	10.7011	78000	0.4020	0.8547
0.4677	10.9754	80000	0.3974	0.8503
0.4482	11.2498	82000	0.3907	0.8410
0.4518	11.5242	84000	0.3890	0.8451
0.4638	11.7986	86000	0.3855	0.8407
0.4325	12.0730	88000	0.3825	0.8328
0.4421	12.3474	90000	0.3742	0.8134
0.4366	12.6218	92000	0.3751	0.8303
0.4412	12.8961	94000	0.3671	0.8120
0.4297	13.1705	96000	0.3639	0.8035
0.4267	13.4449	98000	0.3651	0.8131
0.437	13.7193	100000	0.3575	0.7947
0.4373	13.9937	102000	0.3561	0.8010
0.4182	14.2681	104000	0.3514	0.7860
0.4141	14.5425	106000	0.3490	0.7874
0.4117	14.8168	108000	0.3466	0.7833
0.4116	15.0912	110000	0.3446	0.7753
0.4074	15.3656	112000	0.3411	0.7696
0.4086	15.6400	114000	0.3394	0.7770
0.4189	15.9144	116000	0.3358	0.7666
0.3972	16.1888	118000	0.3332	0.7608
0.3911	16.4632	120000	0.3328	0.7603
0.4062	16.7375	122000	0.3296	0.7496
0.4	17.0119	124000	0.3279	0.7537
0.388	17.2863	126000	0.3262	0.7469
0.3929	17.5607	128000	0.3251	0.7471
0.4036	17.8351	130000	0.3237	0.7455
0.392	18.1095	132000	0.3217	0.7417
0.3916	18.3839	134000	0.3207	0.7406
0.3934	18.6583	136000	0.3190	0.7375
0.3805	18.9326	138000	0.3188	0.7384
0.3778	19.2070	140000	0.3178	0.7365
0.3795	19.4814	142000	0.3173	0.7337
0.3841	19.7558	144000	0.3169	0.7337

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0