hebrew_lyrics_to_singer_classifer_small_dataset

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0080
• Accuracy: 0.9987

Model description

This model was trained to classify singer name based on parts of it's song - on a limited 10 singers dataset of Hebrew lyrics, extracted from Norod78/HebrewLyricsDataet 9 singers_list = {'מתי כספי': 0, 'תיסלם': 1, 'קורין אלאל': 2, 'איפה הילד': 3, 'המכשפות': 4, 'נושאי המגבעת': 5, 'ריקי גל': 6, 'זקני צפת': 7, 'ברי סחרוף': 8, 'אהוד בנאי': 9}

Example use

from transformers import pipeline, AutoModelForSequenceClassification, AutoTokenizer

model_path="yaryar78/hebrew_lyrics_to_singer_classifer_small_dataset"

pipe = pipeline("text-classification", model=model_path)

singers_list = ['מתי כספי' , 'תיסלם', 'קורין אלאל', 'איפה הילד', 'המכשפות', 'נושאי המגבעת', 'ריקי גל', 'זקני צפת', 'ברי סחרוף', 'אהוד בנאי']

singers_list = singers_list[::-1]

label2id = {s: i for i, s in enumerate(singers_list)}

id2label = {i: s for s, i in label2id.items()}

prompt = "ציפורים מסתובבות שמח פה ושמח שם"

prediction = pipe(prompt)

label_str = prediction[0]["label"]

print(label_str)

label_id = int(label_str.replace("LABEL_", ""))

predicted_singer = id2label[label_id]

print(predicted_singer)

---

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: 4
• eval_batch_size: 4
• 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
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7106	1.0	5393	0.1266	0.9643
0.2078	2.0	10786	0.0621	0.9861
0.1516	3.0	16179	0.0246	0.9945
0.1018	4.0	21572	0.0114	0.9978
0.0554	5.0	26965	0.0080	0.9987

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0