README.md

metadata

language:
  - ga
  - en
license: apache-2.0
base_model: openai/whisper-small
tags:
  - generated_from_trainer
datasets:
  - ymoslem/IWSLT2023-GA-EN
  - ymoslem/FLEURS-GA-EN
  - ymoslem/BitesizeIrish-GA-EN
  - ymoslem/SpokenWords-GA-EN-MTed
  - ymoslem/Tatoeba-Speech-Irish
  - ymoslem/Wikimedia-Speech-Irish
metrics:
  - bleu
  - wer
model-index:
  - name: Whisper Small GA-EN Speech Translation
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: IWSLT-2023, FLEURS, BiteSize, SpokenWords, Tatoeba, and Wikimedia
          type: ymoslem/IWSLT2023-GA-EN
        metrics:
          - name: Bleu
            type: bleu
            value: 25.68
          - name: Wer
            type: wer
            value: 71.04907699234579

Whisper Small GA-EN Speech Translation

This model is a fine-tuned version of openai/whisper-small on the IWSLT-2023, FLEURS, BiteSize, SpokenWords, Tatoeba, and Wikimedia dataset. It achieves the following results on the evaluation set:

• Loss: 1.1571
• Bleu: 25.68
• Chrf: 45.53
• Wer: 71.0491

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.03
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Chrf	Wer
2.6685	0.07	100	2.0544	5.05	20.18	139.8919
2.4028	0.13	200	1.7367	12.29	29.72	95.5425
2.1231	0.2	300	1.6141	14.33	30.77	101.3958
1.9192	0.26	400	1.4778	16.86	35.65	91.0851
1.7129	0.33	500	1.3811	16.77	37.53	93.8766
1.5398	0.39	600	1.3427	18.85	39.0	90.2296
1.4257	0.46	700	1.2784	25.73	43.3	70.3287
1.3044	0.53	800	1.2274	25.43	44.33	72.3548
1.2626	0.59	900	1.1875	25.09	44.62	72.6249
1.2801	0.66	1000	1.1571	25.68	45.53	71.0491

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2