You can deploy this model and start earning money today!

MahmutCanBoran/
mbart-audi-diagnosis-agent

Name: mbart-audi-diagnosis-agent
Availability: InStock

facebook/mbart-large-50-many-to-many-mmt

finetune

mit

Views

No views yet

🔧 Audi AI Diagnosis Agent

AI-powered assistant for diagnosing chronic issues in Audi vehicles.
Built with Streamlit, powered by Transformers, and fine-tuned on real-world repair patterns.

🚘 What is this?

Audi AI Diagnosis helps identify possible chronic faults in Audi vehicles.
Using a fine-tuned mBART model, the app turns natural language symptom descriptions into likely diagnoses.

Features

Optimized for Audi-specific problem phrases
Responds in natural, technical language
Real-time inference with Hugging Face Transformers

🚀 Try it out

🧩 Model

This app uses the publicly available model:

🔗 MahmutCanBoran/mbart-audi-diagnosis-agent

Architecture: facebook/mbart-large-50
Task: Text2Text generation (Symptom ➝ Diagnosis)

📦 Dependencies

Create a requirements.txt with:

txt

streamlit>=1.36
transformers>=4.41
torch>=2.2
sentencepiece>=0.2  # Required for mBART tokenization
accelerate>=0.31    # Optional but recommended (for device_map="auto")

💻 Example app.py

python

import streamlit as st
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

MODEL_ID = "MahmutCanBoran/mbart-audi-diagnosis-agent"

@st.cache_resource(show_spinner=True)
def load_model():
    tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
    model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_ID)
    if torch.cuda.is_available():
        model = model.to("cuda")
    return tokenizer, model

st.set_page_config(page_title="Audi AI Diagnosis", page_icon="🚘")
st.title("🚘 Audi AI Diagnosis Agent")
st.caption("mBART-50 based: symptom → likely diagnosis")

symptom = st.text_area(
    "Enter symptom:",
    height=110,
    placeholder="in my A4 40 TDI there is a rattling noise during acceleration"
)

if st.button("Diagnose", type="primary", use_container_width=True):
    tokenizer, model = load_model()
    inputs = tokenizer(symptom.strip(), return_tensors="pt")
    if torch.cuda.is_available():
        inputs = {k: v.to("cuda") for k, v in inputs.items()}
    with torch.inference_mode():
        outputs = model.generate(**inputs, max_new_tokens=96)
    st.success("Likely diagnosis")
    st.write(tokenizer.decode(outputs[0], skip_special_tokens=True))

🧭 Local Setup & Run

# Clone repo
git clone https://huggingface.co/spaces/MahmutCanBoran/audi-ai-diagnosis
cd audi-ai-diagnosis

# Install dependencies
pip install -r requirements.txt

# Run the app
streamlit run app.py