dlgenai-bert-roberta

BERT-base MCQ Answer Selection

Fine-tuned BERT-base-uncased to select the correct answer option (A–E) for multiple-choice questions.

Files

• model.safetensors — fine-tuned weights
• config.json — model configuration
• tokenizer.json / tokenizer_config.json — tokenizer

Training details

• Base model: google-bert/bert-base-uncased
• Task: rank 5 answer options per prompt (top-1 accuracy + mAP@3)
• Format: prompt + 5 options, each option scored via the [CLS] embedding

Usage

from transformers import AutoModelForSequenceClassification, AutoTokenizer
from safetensors.torch import load_file

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained("22f3001251/dlgenai-bert-roberta")

See training/bert_roberta.ipynb in the companion repo for the full training and inference flow.