metadata

language:
  - he
license: apache-2.0
base_model: openai/whisper-medium
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
model-index:
  - name: he-cantillation
    results: []

he-cantillation

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

eval_loss: 0.1666
eval_wer: 10.9128
eval_avg_precision_Exact: 0.9184
eval_avg_recall_Exact: 0.9197
eval_avg_f1_Exact: 0.9188
eval_avg_precision_Letter_Shift: 0.9365
eval_avg_recall_Letter_Shift: 0.9379
eval_avg_f1_Letter_Shift: 0.9369
eval_avg_precision_Word_Level: 0.9382
eval_avg_recall_Word_Level: 0.9395
eval_avg_f1_Word_Level: 0.9385
eval_avg_precision_Word_Shift: 0.9779
eval_avg_recall_Word_Shift: 0.9797
eval_avg_f1_Word_Shift: 0.9784
eval_precision_median_exact: 1.0
eval_recall_median_exact: 1.0
eval_f1_median_exact: 1.0
eval_precision_max_exact: 1.0
eval_recall_max_exact: 1.0
eval_f1_max_exact: 1.0
eval_precision_min_Exact: 0.0
eval_recall_min_Exact: 0.0
eval_f1_min_Exact: 0.0
eval_precision_min_Letter_Shift: 0.0
eval_recall_min_Letter_Shift: 0.0
eval_f1_min_Letter_Shift: 0.0
eval_precision_min_Word_Level: 0.0
eval_recall_min_Word_Level: 0.0
eval_f1_min_Word_Level: 0.0
eval_precision_min_Word_Shift: 0.1429
eval_recall_min_Word_Shift: 0.1111
eval_f1_min_Word_Shift: 0.125
eval_runtime: 1554.5785
eval_samples_per_second: 1.732
eval_steps_per_second: 0.055
epoch: 4.0
step: 50000

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: 8
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: 1000
training_steps: 50000
mixed_precision_training: Native AMP

Framework versions

Transformers 4.39.0.dev0
Pytorch 2.2.1+cu121
Datasets 2.16.1
Tokenizers 0.15.0