smart-mcq-textcnn

Smart MCQ Solver — TextCNN (Trained from Scratch)

Custom TextCNN model for MCQ solving. 5 options (A/B/C/D/E) mein se sahi answer predict karta hai.

Files

• best_model.pt — Trained model weights
• textcnn_config.json — Model configuration
• textcnn_model_code.py — Model class code

How to load

import torch
from transformers import AutoTokenizer
from textcnn_model_code import CustomTextCNN

# Tokenizer
tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
vocab_size = tokenizer.vocab_size

# Model
model = CustomTextCNN(vocab_size=vocab_size)
model.load_state_dict(torch.load("best_model.pt", map_location="cpu"))
model.eval()