patrickvonplaten/wav2vec2-2-bart-base

To rerun this experiment, please clone this directory and run:

python create_model.py

followed by

./run_librispeech.sh

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

wav2vec2-2-bart-base

This model is a fine-tuned version of facebook/wav2vec2-base and bart-base on the librispeech_asr - clean dataset.

It achieves the following results on the evaluation set:

• —Loss: 0.405
• —Wer: 0.0728

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
• —trainbatchsize: 8
• —evalbatchsize: 8
• —seed: 42
• —distributed_type: multi-GPU
• —num_devices: 8
• —totaltrainbatch_size: 64
• —totalevalbatch_size: 64
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —lrschedulerwarmup_steps: 400
• —num_epochs: 5
• —mixedprecisiontraining: Native AMP

Training results

See Training Metrics Tab.

Framework versions

• —Transformers 4.15.0.dev0
• —Pytorch 1.9.0+cu111
• —Datasets 1.16.2.dev0
• —Tokenizers 0.10.3