app.py

import re
import torch
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "magahcicek/avastin-side-effects-model"
model = AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "right"

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

def clean_side_effects(text, drug_name):
    text = text.strip(" ,.\n\t")
    text = re.sub(re.escape(drug_name), '', text, flags=re.IGNORECASE)
    text = re.sub(r"([a-z])([A-Z])", r"\1 \2", text)
    text = re.sub(r"\s*,\s*", ", ", text)
    text = re.sub(r",+", ",", text)
    text = re.sub(r"\s{2,}", " ", text)
    text = re.sub(r"([a-zA-Z])\.([a-zA-Z])", r"\1. \2", text)
    items = re.split(r"[,\n\-]+", text)
    items = [item.strip() for item in items if item.strip()]
    unique_items = []
    for item in items:
        if item.lower() not in [x.lower() for x in unique_items]:
            unique_items.append(item)
    unique_items = [item if item.endswith('.') else item + '.' for item in unique_items]
    return unique_items

def generate_side_effects(drug_name):
    prompt = f"[DRUG_START]{drug_name}[DRUG_END][SIDE_EFFECTS]"
    inputs = tokenizer(prompt, return_tensors="pt", max_length=256, truncation=True, padding="max_length").to(device)
    outputs = model.generate(
        **inputs,
        max_new_tokens=50,
        pad_token_id=tokenizer.eos_token_id,
        num_beams=1,
        repetition_penalty=1.2,
        no_repeat_ngram_size=2,
        early_stopping=True
    )
    full_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return clean_side_effects(full_text, drug_name)

def process_drug_name(drug_name):
    if not drug_name.strip():
        return "Lütfen geçerli bir ilaç adı giriniz."
    side_effects = generate_side_effects(drug_name)
    if isinstance(side_effects, list):
        side_effects_str = "\n- " + "\n- ".join(side_effects)
    else:
        side_effects_str = side_effects
    return f"İlaç Adı: {drug_name}\nÖngörülen Yan Etkiler:\n{side_effects_str}"

iface = gr.Interface(
    fn=process_drug_name,
    inputs=gr.Textbox(lines=1, placeholder="İlaç adını yazınız..."),
    outputs=gr.Textbox(),
    title="Yan Etki Tahmini",
    description="İlaç adını girerek öngörülen yan etkileri liste halinde görün."
)

if __name__ == "__main__":
    iface.launch()