psst_batch_size_4_base_model

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.6743

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss
14.1952	1.68	100	3.6352
3.9092	3.36	200	3.7223
3.9981	5.04	300	3.6864
3.7209	6.72	400	3.6310
3.9395	8.4	500	3.7229
3.7126	10.08	600	3.6163
3.6999	11.76	700	3.6776
3.7203	13.45	800	3.7568
3.7202	15.13	900	3.6998
3.7023	16.81	1000	3.6943
3.689	18.49	1100	3.6501
3.7009	20.17	1200	3.6973
3.6882	21.85	1300	3.6938
3.6907	23.53	1400	3.6795
3.6869	25.21	1500	3.6727
3.681	26.89	1600	3.6749
3.6968	28.57	1700	3.6743

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1
• Datasets 2.12.0
• Tokenizers 0.13.2