hubert-base-superb-ks

This model is a fine-tuned version of facebook/hubert-base-ls960 on the superb dataset. It achieves the following results on the evaluation set:

• Loss: 0.0848
• Accuracy: 0.9822

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: 3e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 0
• distributed_type: IPU
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5.0
• training precision: Mixed Precision

Training results

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.0+cpu
• Datasets 2.1.0
• Tokenizers 0.12.1