app.py

from fastapi import FastAPI
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel, PeftConfig
from mistral_common.tokenizer import AutoMistralTokenizer

# Initialize FastAPI app
app = FastAPI()

# Load PEFT model configuration and base model
config = PeftConfig.from_pretrained("frankmorales2020/Mistral-7B-text-to-sql-flash-attention-2-dataeval")
base_model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")
model = PeftModel.from_pretrained(base_model, "frankmorales2020/Mistral-7B-text-to-sql-flash-attention-2-dataeval")

# Load recommended tokenizer
tokenizer = AutoMistralTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")

# Create the pipeline
from transformers import pipeline
pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer)

@app.get("/")
def home():
    return {"message": "Hello World"}

@app.get("/generate")
def generate(text: str):
    output = pipe(text)
    return {"output": output[0]['generated_text']}