bcash2233/lyrics-generator-gpt2

🎵 AI Lyrics Generator

This model generates song lyrics based on prompts. It's a fine-tuned version of GPT-2 trained on a diverse dataset of song lyrics.

Model Description

• Base Model: GPT-2
• Task: Lyrics Generation
• Training Data: Custom dataset of 10,000+ songs from various genres
• Languages: English

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM

# Load model
tokenizer = AutoTokenizer.from_pretrained("bcash2233/lyrics-generator-gpt2")
model = AutoModelForCausalLM.from_pretrained("bcash2233/lyrics-generator-gpt2")

# Generate lyrics
prompt = "Write lyrics for a rock song about freedom"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(
    inputs.input_ids,
    max_length=300,
    temperature=0.8,
    top_p=0.9,
    do_sample=True
)

lyrics = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(lyrics)

Streamlit Demo

Try the interactive web app:

streamlit run app.py

Training Details

• Epochs: 3
• Batch Size: 4
• Learning Rate: 5e-5
• Training Time: ~2 hours (GPU)

Example Outputs

Prompt: "Write lyrics for a sad ballad about heartbreak"

Output:

I'm standing in the rain again
Thinking 'bout the way we were
Every memory cuts like glass
And I can't seem to let you go...

Limitations

• May generate repetitive phrases
• Best with clear, specific prompts
• Quality varies by prompt complexity

Citation

@misc{lyrics-generator-gpt2,
  author = {Your Name},
  title = {AI Lyrics Generator},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/bcash2233/lyrics-generator-gpt2}
}

License

MIT License