README.md

metadata

library_name: transformers
language:
  - he
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: he-cantillation
    results: []

he-cantillation

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

• Loss: 4.7623
• Wer: 98.8781
• Avg Precision Exact: 0.0291
• Avg Recall Exact: 0.0505
• Avg F1 Exact: 0.0356
• Avg Precision Letter Shift: 0.0493
• Avg Recall Letter Shift: 0.0876
• Avg F1 Letter Shift: 0.0605
• Avg Precision Word Level: 0.0685
• Avg Recall Word Level: 0.1200
• Avg F1 Word Level: 0.0833
• Avg Precision Word Shift: 0.1707
• Avg Recall Word Shift: 0.3059
• Avg F1 Word Shift: 0.2088
• Precision Median Exact: 0.0
• Recall Median Exact: 0.0
• F1 Median Exact: 0.0
• Precision Max Exact: 0.4444
• Recall Max Exact: 1.0
• F1 Max Exact: 0.4286
• Precision Min Exact: 0.0
• Recall Min Exact: 0.0
• F1 Min Exact: 0.0
• Precision Min Letter Shift: 0.0
• Recall Min Letter Shift: 0.0
• F1 Min Letter Shift: 0.0
• Precision Min Word Level: 0.0
• Recall Min Word Level: 0.0
• F1 Min Word Level: 0.0
• Precision Min Word Shift: 0.0
• Recall Min Word Shift: 0.0
• F1 Min Word Shift: 0.0

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: 2
• seed: 42
• optimizer: Use OptimizerNames.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: 1000
• training_steps: 100000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Avg Precision Exact	Avg Recall Exact	Avg F1 Exact	Avg Precision Letter Shift	Avg Recall Letter Shift	Avg F1 Letter Shift	Avg Precision Word Level	Avg Recall Word Level	Avg F1 Word Level	Avg Precision Word Shift	Avg Recall Word Shift	Avg F1 Word Shift	Precision Median Exact	Recall Median Exact	F1 Median Exact	Precision Max Exact	Recall Max Exact	F1 Max Exact	Precision Min Exact	Recall Min Exact	F1 Min Exact	Precision Min Letter Shift	Recall Min Letter Shift	F1 Min Letter Shift	Precision Min Word Level	Recall Min Word Level	F1 Min Word Level	Precision Min Word Shift	Recall Min Word Shift	F1 Min Word Shift
0.0276	3.0030	20000	3.0853	98.7760	0.0336	0.0472	0.0380	0.0581	0.0831	0.0658	0.0801	0.1144	0.0904	0.2032	0.3029	0.2334	0.0213	0.0303	0.0267	0.6667	0.6667	0.6667	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0026	6.0060	40000	3.8257	98.8227	0.0303	0.0497	0.0364	0.0528	0.0880	0.0637	0.0728	0.1208	0.0877	0.1812	0.3095	0.2204	0.0	0.0	0.0	0.75	0.75	0.75	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0004	9.0090	60000	4.3074	99.0124	0.0283	0.0504	0.0347	0.0476	0.0872	0.0590	0.0641	0.1169	0.0792	0.1600	0.2986	0.1987	0.0	0.0	0.0	1.0	1.0	0.8	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0001	12.0120	80000	4.6534	98.8315	0.0295	0.0516	0.0361	0.0502	0.0882	0.0611	0.0687	0.1192	0.0826	0.1715	0.3112	0.2103	0.0	0.0	0.0	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0	15.0150	100000	4.7623	98.8781	0.0291	0.0505	0.0356	0.0493	0.0876	0.0605	0.0685	0.1200	0.0833	0.1707	0.3059	0.2088	0.0	0.0	0.0	0.4444	1.0	0.4286	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu126
• Datasets 2.12.0
• Tokenizers 0.20.1