README.md

---
base_model: facebook/musicgen-small
library_name: transformers
pipeline_tag: text-to-audio
tags:
- lora
- musicgen
- music-generation
- peft
---

# MusicGen LoRA Fine-Tuned

LoRA adapter fine-tuned on custom music dataset using `facebook/musicgen-small`.

## Como usar

```python
from transformers import AutoProcessor, MusicgenForConditionalGeneration
from peft import PeftModel
import torch, soundfile as sf

processor = AutoProcessor.from_pretrained("facebook/musicgen-small")
model = MusicgenForConditionalGeneration.from_pretrained("facebook/musicgen-small")
model.decoder = PeftModel.from_pretrained(model.decoder, "SEU_REPO_AQUI")

inputs = processor(text=["upbeat electronic music"], return_tensors="pt")
audio = model.generate(**inputs, max_new_tokens=256)
sf.write("output.wav", audio[0].cpu().numpy(), 32000)
```