metadata

language:
  - pl
license: apache-2.0
tags:
  - whisper-event
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper Small PL
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          config: pl
          split: test
          args: pl
        metrics:
          - type: wer
            value: 8.589801709229503
            name: Wer
          - type: wer
            value: 8.85
            name: WER
          - type: wer_without_norm
            value: 21.75
            name: WER unnormalized
          - type: cer
            value: 2.63
            name: CER
          - type: mer
            value: 8.76
            name: MER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: facebook/voxpopuli
          type: facebook/voxpopuli
          config: pl
          split: test
        metrics:
          - type: wer
            value: 12.18
            name: WER
          - type: wer_without_norm
            value: 32.17
            name: WER unnormalized

Whisper Small PL

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

Loss: 0.3739
Wer: 8.5898

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: 32
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0474	1.1	1000	0.2561	9.4612
0.0119	3.09	2000	0.2901	8.9726
0.0045	5.08	3000	0.3151	8.8870
0.0007	7.07	4000	0.4218	8.6032
0.0005	9.07	5000	0.3739	8.5898

Evaluation results

When tested on diffrent polish ASR datasets (splits: test), this model achieves the following results:

Dataset	WER	WER unnormalized	CER	MER
common_voice_11_0	??.??	??.??	??.??	??.??

Framework versions

Transformers 4.26.0.dev0
Pytorch 1.13.0+cu117
Datasets 2.7.1.dev0
Tokenizers 0.13.2