devops-bda/abap-api

1import os2from fastapi import FastAPI3from pydantic import BaseModel4from transformers import (5    AutoModelForCausalLM, 6    AutoTokenizer, 7    pipeline, 8    AutoConfig9)10 11# Load the configuration and remove any quantization config if present12config = AutoConfig.from_pretrained("devops-bda/Abap")13if hasattr(config, "quantization_config"):14    del config.quantization_config  # Safely delete it instead of setting to None15 16# Load the model and tokenizer without 4-bit quantization17model = AutoModelForCausalLM.from_pretrained(18    "devops-bda/Abap",19    config=config,20    device_map="auto"  # This ensures the model loads properly on CPU21)22tokenizer = AutoTokenizer.from_pretrained("devops-bda/Abap")23 24# Create a text-generation pipeline with the loaded model and tokenizer25text_gen_pipeline = pipeline("text-generation", model=model, tokenizer=tokenizer)26 27app = FastAPI()28 29class InputData(BaseModel):30    input_text: str31 32@app.get("/health")33async def health_check():34    return {"status": "ok", "message": "Model is ready"}35 36@app.post("/predict")37async def predict(data: InputData):38    output = text_gen_pipeline(data.input_text, max_length=500)39    return {"output": output}