README.md

---
license: mit
datasets:
- Miguelpef/3d-prompt
language:
- es
base_model:
- facebook/bart-base
new_version: Miguelpef/bart-base-lora-3DPrompt
pipeline_tag: text-generation
library_name: transformers
tags:
- 3d
- prompt
- español
---

![Miguelpef/bart-base-lora-3DPrompt](images/ModeloLora.jpg)

Spanish version

**The model is still in the training phase. This is not the final version and may contain artifacts and perform poorly in some cases.**

## Setting Up
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from peft import PeftModel, PeftConfig

# Define the repository ID
repo_id = "Miguelpef/bart-base-lora-3DPrompt"

# Load the PEFT configuration from the Hub
peft_config = PeftConfig.from_pretrained(repo_id)

# Load the base model from the Hub
model = AutoModelForSeq2SeqLM.from_pretrained(peft_config.base_model_name_or_path)

# Load the tokenizer from the Hub
tokenizer = AutoTokenizer.from_pretrained(repo_id)

# Wrap the base model with PEFT
model = PeftModel.from_pretrained(model, repo_id)

# Now you can use the model for inference as before
def generar_prompt_desde_objeto(objeto):
    prompt = objeto
    inputs = tokenizer(prompt, return_tensors='pt').to(model.device)
    outputs = model.generate(**inputs, max_length=100)
    prompt_generado = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return prompt_generado

mi_objeto = "Mesa grande marrón" #Change this object
prompt_generado = generar_prompt_desde_objeto(mi_objeto)
print({prompt_generado})
```