metadata
library_name: peft
language:
- en
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- wft
- whisper
- automatic-speech-recognition
- audio
- speech
- generated_from_trainer
datasets:
- hf-internal-testing/librispeech_asr_dummy
metrics:
- wer
model-index:
- name: wft-test-model
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: hf-internal-testing/librispeech_asr_dummy
type: hf-internal-testing/librispeech_asr_dummy
metrics:
- type: wer
value: 65.4320987654321
name: Wer
wft-test-model
This model is a fine-tuned version of openai/whisper-tiny on the hf-internal-testing/librispeech_asr_dummy dataset. It achieves the following results on the evaluation set:
- Loss: 1.0315
- Wer: 65.4321
- Cer: 30.1947
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: 0.0005
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 1.8721 | 1.0 | 8 | 1.5609 | 74.0741 | 31.3264 |
| 0.8099 | 2.0 | 16 | 1.2784 | 69.6296 | 31.1453 |
| 0.3841 | 3.0 | 24 | 1.1096 | 75.3086 | 36.5776 |
| 0.1898 | 4.0 | 32 | 1.0460 | 68.6420 | 32.5939 |
| 0.1511 | 5.0 | 40 | 1.0315 | 65.4321 | 30.1947 |
Framework versions
- PEFT 0.13.2
- Transformers 4.45.2
- Pytorch 2.5.0
- Datasets 3.0.2
- Tokenizers 0.20.1