app.py

import torch
import gradio as gr
import pandas as pd
from datasets import Dataset
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    TrainingArguments,
    Trainer
)
import os

token = os.getenv('HF_TOKEN')

# load dataset
df = pd.read_csv("dataset.csv")
dataset = Dataset.from_pandas(df)

# load tokenizer & model
model_name = "MoritzLaurer/mDeBERTa-v3-base-mnli-xnli"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=2, ignore_mismatched_sizes=True)

# tokenize data
def preprocess(examples):
    return tokenizer(examples["text"], truncation=True, padding=True)

tokenized_dataset = dataset.map(preprocess, batched=True)

# training arguments
training_args = TrainingArguments(
    output_dir="./results",
    per_device_train_batch_size=4,
    num_train_epochs=3,
    logging_steps=10,
    save_strategy="no",
    learning_rate=2e-5,
)

# train
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_dataset,
    tokenizer=tokenizer,
)
trainer.train()

from huggingface_hub import HfApi

model.push_to_hub("kwanpon/mdeberta-classify-thai", use_auth_token=token)
tokenizer.push_to_hub("kwanpon/mdeberta-classify-thai", use_auth_token=token)

# inference function for gradio
def classify(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
    with torch.no_grad():
        outputs = model(**inputs)
    probs = torch.softmax(outputs.logits, dim=1).numpy()[0]
    return {
        "ไม่เกี่ยวข้อง": float(probs[0]),
        "จ้างงานรถขนส่ง": float(probs[1]),
    }

# gradio interface
demo = gr.Interface(
    fn=classify,
    inputs=gr.Textbox(lines=3, label="ข้อความ"),
    outputs=gr.Label(label="ผลการจำแนก"),
    title="Text Classifier: Zero-Shot NLI",
    description="กรุณาพิมพ์ข้อความเพื่อตรวจสอบว่าเป็นการว่าจ้างงานรถขนส่งหรือไม่"
)

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