from fastapi import FastAPI, Form
from fastapi.responses import HTMLResponse
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
import torch
app = FastAPI()
MODEL_ID = "ibm-granite/granite-4.0-tiny-preview"
# Load tokenizer and model
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(
MODEL_ID,
torch_dtype=torch.float16 if torch.cuda.is_available() else "auto",
device_map="auto"
)
# Use pipeline for easier text generation (no device arg when using device_map="auto")
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
@app.get("/", response_class=HTMLResponse)
def index():
return """
Granite Tiny Summarizer
Granite 4.0 Tiny Summarization Demo
"""
@app.post("/summarize", response_class=HTMLResponse)
def summarize(text: str = Form(...)):
prompt = (
"Below is a passage of text. Please provide a concise summary in 2-4 sentences.\n\n"
f"Text:\n{text.strip()}\n\nSummary:"
)
outputs = pipe(
prompt,
max_new_tokens=150,
do_sample=True,
temperature=0.7,
top_p=0.95,
eos_token_id=tokenizer.eos_token_id,
pad_token_id=tokenizer.eos_token_id
)
output_text = outputs[0]['generated_text']
summary = output_text.split("Summary:")[-1].strip()
return f"Summary
{summary}
Back"