controngo's picture
Upload tokenizer
fe204ca verified
|
raw
history blame
2.07 kB
metadata
base_model: openai/whisper-tiny
license: apache-2.0
metrics:
  - wer
tags:
  - generated_from_trainer
model-index:
  - name: whisper-tinyfinacial
    results: []

whisper-tinyfinacial

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

  • Loss: 0.5217
  • Wer: 55.6180

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: 1.35e-05
  • train_batch_size: 8
  • 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: 100
  • training_steps: 600

Training results

Training Loss Epoch Step Validation Loss Wer
No log 0.4 50 0.9091 64.0449
No log 0.8 100 0.6941 52.2472
No log 1.2 150 0.5615 51.6854
No log 1.6 200 0.5219 47.1910
No log 2.0 250 0.4938 47.7528
No log 2.4 300 0.4970 50.0
No log 2.8 350 0.4999 58.4270
No log 3.2 400 0.5076 46.0674
No log 3.6 450 0.5157 52.2472
0.3104 4.0 500 0.5277 56.1798
0.3104 4.4 550 0.5257 57.3034
0.3104 4.8 600 0.5217 55.6180

Framework versions

  • Transformers 4.41.2
  • Pytorch 2.3.0+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1