metadata

base_model: openai/whisper-small
library_name: transformers
license: apache-2.0
metrics:
  - wer
tags:
  - generated_from_trainer
model-index:
  - name: names-whisper-en-spectrogram-pitch-shifted
    results: []

names-whisper-en-spectrogram-pitch-shifted

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.2384
Ner percent: 101.9340
Wer: 5.3520

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: 16
eval_batch_size: 8
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	Ner percent	Wer
0.036	5.0251	1000	0.2288	99.9682	6.2317
0.0013	10.0503	2000	0.2292	97.7273	5.6542
0.0005	15.0754	3000	0.2334	102.4688	5.3084
0.0003	20.1005	4000	0.2368	101.9340	5.3621
0.0003	25.1256	5000	0.2384	101.9340	5.3520

Framework versions

Transformers 4.44.2
Pytorch 2.3.0+cu121
Datasets 2.19.0
Tokenizers 0.19.1