README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: kids_phoneme_sm_model
    results: []

kids_phoneme_sm_model

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: 15.8911
• Cer: 1.0

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: 4e-59
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
6.5005	0.74	500	15.8911	1.0
6.3865	1.48	1000	15.8911	1.0
6.3822	2.22	1500	15.8911	1.0
6.4251	2.96	2000	15.8911	1.0
6.3836	3.7	2500	15.8911	1.0
6.4552	4.44	3000	15.8911	1.0
6.4287	5.19	3500	15.8911	1.0
6.4304	5.93	4000	15.8911	1.0
6.375	6.67	4500	15.8911	1.0
6.5434	7.41	5000	15.8911	1.0
6.319	8.15	5500	15.8911	1.0
6.4693	8.89	6000	15.8911	1.0
6.4585	9.63	6500	15.8911	1.0
6.4795	10.37	7000	15.8911	1.0
6.393	11.11	7500	15.8911	1.0
6.3489	11.85	8000	15.8911	1.0
6.4511	12.59	8500	15.8911	1.0
6.3378	13.33	9000	15.8911	1.0
6.5094	14.07	9500	15.8911	1.0
6.3643	14.81	10000	15.8911	1.0
6.5079	15.56	10500	15.8911	1.0
6.4272	16.3	11000	15.8911	1.0
6.3937	17.04	11500	15.8911	1.0
6.3704	17.78	12000	15.8911	1.0
6.3797	18.52	12500	15.8911	1.0
6.5126	19.26	13000	15.8911	1.0
6.4111	20.0	13500	15.8911	1.0
6.3557	20.74	14000	15.8911	1.0
6.5705	21.48	14500	15.8911	1.0
6.3207	22.22	15000	15.8911	1.0
6.4664	22.96	15500	15.8911	1.0
6.4401	23.7	16000	15.8911	1.0
6.4136	24.44	16500	15.8911	1.0
6.5345	25.19	17000	15.8911	1.0
6.4097	25.93	17500	15.8911	1.0
6.2736	26.67	18000	15.8911	1.0
6.5474	27.41	18500	15.8911	1.0
6.3997	28.15	19000	15.8911	1.0
6.3997	28.89	19500	15.8911	1.0
6.5538	29.63	20000	15.8911	1.0

Framework versions

• Transformers 4.30.1
• Pytorch 2.0.0
• Datasets 2.12.0
• Tokenizers 0.13.3