finetune_wav2vec2_base_six_500

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.9216
• Wer: 100.0
• Cer: 99.5843

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: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• training_steps: 4500

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
4.1782	9.2593	500	2.9234	100.0	99.5862
2.0275	18.5185	1000	1.2454	57.1029	25.9689
0.565	27.7778	1500	1.1227	47.9586	22.3906
0.3036	37.0370	2000	1.2474	43.9597	21.5689
0.1954	46.2963	2500	1.3963	42.9251	20.9569
0.1336	55.5556	3000	1.5303	43.4843	21.5514
0.0999	64.8148	3500	1.4380	42.2260	20.7355
0.0797	74.0741	4000	1.5207	41.4430	20.6422
0.0648	83.3333	4500	1.5408	41.0235	20.4557

Framework versions

• Transformers 4.40.2
• Pytorch 1.12.1+cu116
• Datasets 2.19.1
• Tokenizers 0.19.1