metadata

library_name: transformers
license: apache-2.0
base_model: openai/whisper-medium
tags:
  - generated_from_trainer
datasets:
  - swagen
metrics:
  - wer
model-index:
  - name: whisper-medium-swagen-combined-25hrs-model
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: swagen
          type: swagen
        metrics:
          - name: Wer
            type: wer
            value: 0.25892857142857145

whisper-medium-swagen-combined-25hrs-model

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

Loss: 0.3662
Wer: 0.2589

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: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 30.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.8233	0.0993	200	0.8047	0.4897
1.9329	0.1986	400	0.6191	0.4011
1.6927	0.2980	600	0.5421	0.3791
1.6183	0.3973	800	0.4889	0.3210
1.4431	0.4966	1000	0.4684	0.2866
1.4117	0.5959	1200	0.4258	0.2650
1.2699	0.6952	1400	0.4222	0.2665
1.0532	0.7945	1600	0.4108	0.2513
1.0589	0.8939	1800	0.3982	0.2291
1.1856	0.9932	2000	0.3853	0.2355
0.6692	1.0929	2200	0.4001	0.2650
0.6505	1.1922	2400	0.3919	0.2389
0.6613	1.2915	2600	0.3809	0.2385
0.6194	1.3908	2800	0.3873	0.2343
0.6358	1.4901	3000	0.3850	0.2142
0.6208	1.5894	3200	0.3779	0.2388
0.5932	1.6888	3400	0.3725	0.2040
0.5797	1.7881	3600	0.3712	0.2092
0.5707	1.8874	3800	0.3738	0.2342
0.5928	1.9867	4000	0.3662	0.2589
0.2626	2.0864	4200	0.3803	0.2697
0.2557	2.1857	4400	0.3853	0.2102
0.3342	2.2850	4600	0.3891	0.2062

Framework versions

Transformers 4.47.1
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0