Osiris_asr_model – AI Model by tiaTai | AlphaNeural AI

You can deploy this model and start earning money today!

tiaTai/
Osiris_asr_model

like

0

automatic-speech-recognition

generated_from_trainer

facebook/wav2vec2-base

endpoints_compatible

Views

No views yet

Osiris_asr_model

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: 3.0600
Wer: 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: 1e-05
train_batch_size: 10
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 20
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 50
training_steps: 1000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
45.8209	50.0	50	21.0347	1.0182
10.2898	100.0	100	5.1552	1.0
5.7188	150.0	150	4.9140	1.0
5.3358	200.0	200	4.7650	1.0
5.1381	250.0	250	4.6797	1.0
4.9841	300.0	300	4.6168	1.0
4.9255	350.0	350	4.5741	1.0
4.8353	400.0	400	4.5321	1.0
4.7704	450.0	450	4.5100	1.0
4.6257	500.0	500	3.9382	1.0
3.8106	550.0	550	3.3939	1.0
3.5095	600.0	600	3.2887	1.0
3.3716	650.0	650	3.1967	1.0
3.3025	700.0	700	3.1539	1.0
3.2532	750.0	750	3.1477	1.0
3.2086	800.0	800	3.0984	1.0
3.1889	850.0	850	3.0857	1.0
3.162	900.0	900	3.0819	1.0
3.1411	950.0	950	3.0610	1.0
3.1397	1000.0	1000	3.0600	1.0

Framework versions

Transformers 4.35.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.0